FORSCHUNGSBERICHT AGRARTECHNIK des Fachausschusses Forschung und Lehre der Max-Eyth-Gesellschaft Agrartechnik im VDI (VDI-MEG) 598

Carolina Bilibio

Evapotranspiration and Drainage of Potash Tailings Covers

Dissertation Witzenhausen 2018

Universität Kassel Fachbereich Ökologische Agrarwissenschaften Fachgebiet Agrartechnik Prof. Dr. sc. agr. Oliver Hensel

Evapotranspiration and Drainage of Potash Tailings Covers

Dissertation zur Erlangung des akademischen Grades Doktor der Agrarwissenschaften (Dr. agr.)

von Carolina Bilibio aus Santo Augusto, Rio Grande do Sul, Brasilien 2018

Die vorliegende Arbeit wurde vom Fachbereich für Ökologische Agrarwissenschaften, Fachgebiet Agrartechnik der Universität Kassel als Dissertation zur Erlangung des akademischen „Grades Doktor der Agrarwissenschaften“ angenommen.

Tag der mündlichen Prüfung: 12.07.2018

Erster Gutachter: Prof. Dr. Oliver Hensel Zweiter Gutachter: Prof. Dr. Helge Schmeisky

Mündliche Prüfer: Prof. Dr. sc. agr. Oliver Hensel Prof. Dr. Helge Schmeisky Prof. Dr. Stephan Peth Dr. Christian Bruns

Alle Rechte vorbehalten. Die Verwendung von Texten und Bildern, auch auszugsweise, ist ohne Zustimmung des Autors urheberrechtswidrig und strafbar. Das gilt insbesondere für Vervielfältigung, Übersetzung, Mikroverfilmung sowie die Einspeicherung und Verarbeitung in elektronischen Systemen. © 2018 Im Selbstverlag: Carolina Bilibio Bezugsquelle: Universität Kassel, FB Ökologische Agrarwissenschaften Fachgebiet Agrartechnik Nordbahnhofstr. 1a 37213 Witzenhausen

Dedication

Dedication

This work is dedicated with love and gratitude to my parents, Valdemar Bilibio (in memorian) and Clarice Lúcia Bilibio

Este trabalho é dedicado com amor e gratidão aos meus pais, Valdemar Bilibio (in memorian) e Clarice Lúcia Bilibio

Affidaviti

Affidavit

I herewith give assurance that I completed this dissertation independently without prohibited assistance of third parties or aids other than those identified in this dissertation. All passages that are drawn from published or un-published writings, either word-for-word or in paraphrase, have been clearly identified as such. Third parties were not involved in the drafting of the material content of this dissertation; most specifically I did not employ the assistance of a dissertation advisor. No part of this thesis has been used in another doctoral or tenure process.

Erklärung

Hiermit versichere ich, dass ich die vorliegende Dissertation selbständig, ohne unerlaubte Hilfe Dritter angefertigt und andere als die in der Dissertation angegebenen Hilfsmittel nicht benutzt habe. Alle Stellen, die wörtlich oder sinngemäß aus veröffentlichten oder unveröffentlichten Schriften entnommen sind, habe ich als solche kenntlich gemacht. Dritte waren an der inhaltlichen Erstellung der Dissertation nicht beteiligt; insbesondere habe ich nicht die Hilfe eines kommerziellen Promotionsberaters in Anspruch genommen. Kein Teil dieser Arbeit ist in einem anderen Promotions- oder Habilitationsverfahren durch mich verwendet worden.

Witzenhausen, den 12.07.2018 Carolina Bilibio

Acknowledgements

Acknowledgements

This study was supported by K+S KALI GmbH, project number 6525103, with the title: “Greening Concept for Seepage Reduction from Potash Tailings Piles in Region - Stage 2 / Lysimeter Experiment at Pile IV in ”.

I thank:

 my supervisor, Prof. Dr. Oliver Hensel, for the opportunity, support and guidance to complete my doctorate in Germany. I am grateful to have worked on this project. I also thank my supervisor for funding the physical-hydraulic measurements of the substrates and for allowing me to participate in the Hydrus course in the Netherlands.

 my second supervisor, Prof. Dr. Helge Schmeisky, for the experimental observations and suggestions; the thesis committee, Prof. Dr. Stephan Peth and Dr. Christian Bruns, for the scientific comments and discussions.

 Prof. Dr. Ir. Rien van Genuchten for the teaching and recommendations during the Hydrus course in the Netherlands. I also thank Prof. Dr. Ir. Rien van Genuchten for the reviews and comments for my publications.

 Dr. Daniel Uteau Pushmann and Mrs. Margit Rode for the scientific and technical assistance in the soil laboratories; Philipp Tony Ruttimann for the English proofreading; and the Elsevier author´s center for polishing the graphical abstracts and pictures.

 the colleagues from the Department of Agricultural and Biosystems Engineering, Dr. Stefanie Retz and Christian Schellert for advice and training in the laboratory. Greta Papke, Heiko Tostmann, Paula Brenner, Matthias Brütting, Johanna Hoppe, and Max Ahlert, for the help with samplings.

I am very grateful for the encouragement and support I have received from my family: my mother Clarice, my sisters Elisabete and Cristina, my brothers José and Augusto; my aunt, Sister Anair Segala, my friend Maria Elisabeth Goebel, and my boyfriend Hans-Hermann Kaufmann, that made it possible to complete this work. I thank especially Augusto Valdemar Bilibio and Elisabete Bilibio for the support during my stay in Germany.

Eu agradeço o apoio e encorajamento que recebi da minha Mãe Clarice, minhas manas Elisabete e Cristina, meus irmãos José e Augusto, minha Tia Irmã Anair, minha amiga Maria, e meu namorado Hans-Hermann, que tornou possível concluir este trabalho. Agradeço em especial o Augusto e a Elisabete pelo suporte durante a minha estadia na Alemanha.

Preliminary remarks

Preliminary remarks

The following chapters of this thesis are either included in publications or prepared to be submitted in scientific journals.

Chapter 3: Bilibio, C., Schellert, C., Retz, S., Hensel, O., Schmeisky, H., Uteau, D., Peth, S., 2017. Water balance assessment of different substrates on potash tailings piles using non-weighable lysimeters. J. Environ. Manage. 196, 633-643. Doi: 10.1016/j.jenvman.2017.01.024

Chapter 4: Bilibio, C., Hensel, O., van Genuchten, R., Uteau, D., Peth, S., 2018. Simulation of evapotranspiration and drainage on potash tailings covers using Hydrus-1D. Vadose Zone Journal (Draft manuscript to be submitted).

Chapter 5: Bilibio, C., Hensel, O., 2018. The water deficit of evapotranspiration covers on potash tailing piles using CropWat. Agricultural and Forest Meteorology (Draft manuscript to be submitted).

Table of contents

Table of contents List of figures ...... iv List of tables ...... viii List of symbols ...... xii 1 General introduction ...... 1 1.1 Objectives and hypotheses of the research ...... 3 1.2 Thesis structure ...... 4 1.3 References ...... 4 2 State of the art ...... 8 2.1 Water cycle ...... 8 2.1.1 Precipitation ...... 8 2.1.2 Infiltration ...... 9 2.1.3 Redistribution ...... 11 2.1.4 Runoff ...... 11 2.1.5 Evapotranspiration ...... 12 2.2 Mining and potash tailings...... 50 2.3 Evapotranspiration covers and technogenic substrates ...... 52 2.3.1 Perennial grasses ...... 55 2.3.2 Waste system in Germany ...... 56 2.3.3 Mathematical models for evapotranspiration covers ...... 57 2.4 References ...... 58 3 Water Balance Assessment of Different Substrates on Potash Tailings Piles using Non- Weighable Lysimeters ...... 70 3.1 Graphical abstract...... 70 3.2 Highlights ...... 70 3.3 Abstract...... 71 3.4 Introduction ...... 72 3.5 Materials and methods ...... 75 3.5.1 Description of the experimental area ...... 75 3.5.2 Experimental design and installation of lysimeters ...... 76 3.5.3 Drainage ...... 78 3.5.4 Seeding and fertilization ...... 79 3.5.5 Meteorological data ...... 79 3.5.6 Determination of water balance components ...... 80 3.5.7 Reference crop evapotranspiration ...... 80

i

Table of contents

3.5.8 Evaluation period and statistical analysis ...... 81 3.6 Results and discussion ...... 82 3.6.1 Weather condition...... 82 3.6.2 Water balance ...... 87 3.7 Implications and limitations of the study ...... 91 3.8 Conclusions...... 92 3.9 References ...... 93 4 Simulation of Evapotranspiration and Drainage on Potash Tailings Covers using Hydrus-1D 101 4.1 Graphical abstract...... 101 4.2 Highlights ...... 101 4.3 Abstract...... 102 4.4 Introduction ...... 103 4.5 Material and methods ...... 104 4.5.1 Experimental site and design ...... 105 4.5.2 Meteorological data ...... 106 4.5.3 Drainage and evapotranspiration assessment ...... 109 4.5.4 Hydraulic properties, pH and electrical conductivity of the substrates ...... 110 4.5.5 Simulations and calibration of Hydrus-1D ...... 113 4.5.6 Evapotranspiration and drainage of different rates of fine fractions and soil textures 119 4.5.7 Statistical Analyses ...... 120 4.6 Results and discussions ...... 120 4.6.1 Hydraulic parameters of the substrates ...... 120 4.6.2 Observed water fluxes of the substrates ...... 129 4.6.3 Forward simulation, calibration and validation of the Hydrus-1D model ...... 131 4.6.4 Validation of the calibrated Hydrus-1D model and predictions ...... 137 4.6.5 Water fluxes using different rates of fine fractions and soil textures ...... 139 4.7 Conclusions...... 143 4.8 References ...... 144 4.9 Supplementary materials ...... 149 5 The Water Deficit of Evapotranspiration Covers on Potash Tailing Piles Using CropWat...... 176 5.1 Graphical abstract...... 176 5.2 Highlights ...... 176 5.3 Abstract...... 177 5.4 Introduction ...... 178 5.5 Material and methods ...... 179 5.5.1 Experimental site and design ...... 179 ii

Table of contents

5.5.2 Meteorological data ...... 181 5.5.3 Drainage and evapotranspiration assessment ...... 181 5.5.4 Seeding and fertilization ...... 181 5.5.6 CropWat configuration ...... 182 5.5.7 Further simulations ...... 186 5.5.8 Statistical analyses ...... 187 5.6 Results and discussions ...... 187 5.6.1 Weather data ...... 187 5.6.2 Crop evapotranspiration and water deficit ...... 189 5.6.3 Further simulations ...... 198 5.7 Conclusions...... 205 5.8 References ...... 205 6 General discussion ...... 209 6.1 References ...... 214 7 General conclusions ...... 216 8 Summary ...... 218 9 Appendix ...... 224

iii

List of figures

List of figures

Figure 3-1: Part of the potash tailings pile named “Monte Kali” from the Wintershall production facility which belongs to K+S KALI GmbH. The pile is in the outskirts of Heringen, Germany. The experimental site was situated in the proximities of the conveyor belts which transport the tailings from the processing facilities to the heap (Picture from K+S KALI GmbH)...... 75 Figure 3-2: An example of the non- weighable lysimeter with 3-m deep and covering an area of 2 m2. The surface of the lysimeter included the substrate mixture plus organic compost. The subsurface layer contained the different portions of household waste incineration slags and coal combustion residues. Whereas the filter layer incorporated different fractions of gravel in order to avoid the washing out of the substrates...... 77 Figure 3-3: Ombrometer from Thies weather station (a), ground-level rain gauge (b) and 1-m-high rain gauge (c) ...... 79 Figure 3-4: Monthly and biennial mean values of the mean air temperature, substrates temperature, relative air humidity, solar radiation, wind speed and reference evapotranspiration from 2014 to 2015...... 85 Figure 3-5: Water balance components of different substrates during winter and summer for 2014 and 2015 (Data are total sum ± standard deviation). Ground-level rain gauges n=4.

Drainage n=2. Where ETa: actual evapotranspiration; ET0: reference crop

evapotranspiration; ETc: crop evapotranspiration; ETH: Haude´s evapotranspiration 90 Figure 4-1: Aerial view of potash tailings from Wintershall potash plant on 14 May 2012. Size of the picture: 1500 pixels in horizontal versus 1125 pixels in vertical image. The horizontal ground sampling distance, GSD, is 1.4462 m/pixel and the vertical GSD is 1.4460 m/pixel (TerraServer, 2016) ...... 105 Figure 4-2: Distribution of the lysimeters and treatments at the experiment field...... 106 Figure 4-3: Daily minimum (gray line) and maximum (black line) air temperature, relative air humidity, solar radiation and wind speed (2-m height) registered at the experimental site during three hydrological years...... 108 Figure 4-4: (a) Material distribution; (b) root distribution; (c) initial pressure head; (d) observation points for the forward simulation using Hydrus-1D ...... 117 Figure 4-5: pH values of the substrates 1-4 in 2014 and in 2016 according to different depths .. 121 Figure 4-6: Electrical conductivity of substrates 1-4 in 2014 and in 2016 according to different depths ...... 122 Figure 4-7: Bulk density and total porosity of the substrates 1-4 in 2016 according to different depths ...... 123

iv

List of figures

Figure 4-8: Water retention curve from substrate 1 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6)...... 124 Figure 4-9: (a) Mean saturated hydraulic conductivity for each substrate and depth (n=6); (b) mean saturated hydraulic conductivity for the substrates 1-4 at different depths (n=24) ...... 125 Figure 4-10: Accumulated infiltration of substrates 1, 2 and 4 in 2014 according to different depths ...... 126 Figure 4-11: (a) Cumulative grain size distribution curves of fine and coarse particles from substrates 1-4; (b) Cumulative grain size distribution curves of fine particles from substrates 1-4. Mean values from 0.0 to 3.0 m depth...... 127 Figure 4-12: Weekly water content of substrates 1-4 from 25.06.2015 to 27.10.2016 ...... 129 Figure 4-13: Observed water fluxes of substrates 1-4 during 2014, 2015 and 2016 (± standard deviation)...... 130 Figure 4-14: Observed and optimized water retention curve from substrates 1-4. The solid line is the observed model from 0.0 to 0.64 m depth. The dashed line is the optimized model from 0.0 to 2.60 m depth using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D over two hydrological years ... 133 Figure 4-15: Observed versus predicted accumulated drainage of substrates 1-4 using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D during two hydrological years ...... 134 Figure 4-16: Predicted potential and actual root water uptake of substrate 1 using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D over two hydrological years ...... 135 Figure 4-17: Predicted water content levels in different observation points of substrate 1 using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D during two hydrological years ...... 136 Figure 4-18: Predicted water storage of the substrates 1-4 in the entire flow domain of the lysimeters, from 0.0 to 2.6 m depth ...... 136 Figure 4-19: Accumulated water fluxes for 27 water-years for Bad Hersfeld using optimized hydraulic parameters from substrate 1 ...... 138 Figure 4-20: Seepage rates according to different root depths and crop height for 27 water-years for Bad Hersfeld using optimized hydraulic parameters from substrates 1 ...... 139 Figure 4-21: Dry bulk density of the substrates 1-4 according to different rates of fine fraction (± standard deviation) ...... 140

v

List of figures

Figure 4-22: Total porosity of the substrates 1-4 according to different rates of fine fractions (± standard deviation) ...... 140 Figure 4-23: (a) Water retention curve using 60, 80 and 100 % of fine particles using texture and bulk densities measurements at the Rosetta pedotransfer function; (b) water retention curve according to the relative stone content for substrate 1 ...... 142 Figure 4-24: (a) Water retention curve from substrate 2 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6) ...... 150 Figure 4-25: (a) Water retention curve from substrate 3 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6) ...... 150 Figure 4-26: (a) Water retention curve from substrate 4 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6) ...... 151 Figure 5-1: (a) Monte Kali in the outskirts of the Hessian city of Heringen, (b) Germany ...... 180 Figure 5-2: Water retention curve of the substrates 1-4 from 0.0 to 0.64 m depth ...... 184 Figure 5-3: Minimum and maximum air temperature, relative air humidity, sun hours, wind speed and precipitation in the lysimeter experimental site during 3 calendar years ...... 188 Figure 5-4: Effective precipitation, crop evapotranspiration and water deficit for substrates 1-4 during 2014, 2015 and 2016. Values within the columns refer to the water deficit of

the respective month (mm). Water deficit = Peff. - ETc (10-days interval) ...... 189 Figure 5-5: Crop evapotranspiration and actual evapotranspiration under rainfed conditions for substrates 1-4 in 2014, 2015 and 2016 ...... 191

Figure 5-6: Soil moisture depletion under optimum (ETc) and rainfed (ETa) condition for

substrates 1-4 during 2014, 2015 and 2016. θcc: field capacity ...... 192

Figure 5-7: Water stress coefficient under optimum (Ks ETc) and rainfed (Ks ETa) conditions for

substrate 1 in 2014, 2015 and 2016. θcc: field capacity. θpmp: permanent wilting point...... 193 Figure 5-8: (a) Probability of exceedance of annual precipitation, points are the annual precipitations registered in Bad Hersfeld from 1987 to 2016 (30 calendar years). (b) Monthly precipitation according to different degrees of probability...... 199 Figure 5-9: (a) Minimum and maximum air temperature, sun hours; (b) relative air humidity and wind speed; in Bad Hersfeld from 1987 to 2016 (30 years) ...... 199 Figure 5-10: Effective precipitation, crop evapotranspiration and water deficit for a green cover under different precipitation probabilities, 20, 50 and 80 %, in substrate 1 ...... 200

vi

List of figures

Figure 5-11: Daily crop evapotranspiration and actual evapotranspiration for a green cover under different exceedance probabilities of the precipitation, 20, 50 and 80 % and substrate 1 ...... 202

Figure 5-12: Soil moisture depletion for a green cover under optimum (ETc SMD) and rainfed

conditions (ETa SMD) considering different exceedance probabilities of the

precipitation and substrate 1. θcc: field capacity ...... 202 Figure 5-13: Effective precipitation, crop evapotranspiration and water deficit for a green cover under normal precipitation depths and different crop coefficients in substrate 1 ..... 203 Figure 5-14: Crop evapotranspiration and actual evapotranspiration of a green cover under normal precipitation depths and different crop coefficients in substrate 1 ...... 204

Figure 5-15: Soil moisture depletion under optimum (ETc) and rainfed (ETa) condition for normal

precipitation depths and different crop coefficients in substrate 1. θcc: field capacity ...... 204 Figure 6-1: Actual evapotranspiration and seepage depth during winter and summer from 2014 to 2016 for substrates 1-4. ± Standard deviation ...... 211 Figure 9-1: Photo documentation of the lysimeter experimental site in Heringen (Werra) ...... 224 Figure 9-2: Photo documentation of the laboratory measurements in the Agricultural and Biosystems Engineering and Soil Science laboratories of the University of Kassel 229

vii

List of tables

List of tables

Table 2-1: Measurements and mathematical expressions to estimate reference evapotranspiration ...... 42 Table 3-1: Water retention curve parameters according to van Genuchten (1980) model of the substrates 1 and 4 ...... 78 Table 3-2: Water volume (%) released by gravity, available and unavailable moisture in substrates 1 and 4 ...... 78 Table 3-3: Precipitation (mm) at the Heringen experimental site during two hydrological years and climatological normals ...... 82 Table 3-4: Drainage of lysimeters according to different substrates at the Heringen experimental site during the two hydrological years ...... 87 Table 3-5: Actual evapotranspiration (mm) according to different substrates at the Heringen experimental site during two hydrological years ...... 89 Table 4-1: Meteorological data of the Heringen experimental field during three hydrological years ...... 107 Table 4-2: Hydrus 1-D inputs for the water balance simulation ...... 114 Table 4-3: Inverse solution data for the Hydrus 1-D calibration ...... 117

Table 4-4: Median equivalent diameter, d50, and uniformity coefficients of substrates 1-4 for the entire substrates (diameter < 12 mm) and for the fines (diameter < 2 mm) ...... 128 Table 4-5: Mean hydraulic parameters and accumulated seepage using 60%, 80% and 100% fine fraction of substrates 1-4 and soil textures from 2014 to 2015 in Heringen ...... 141 Table 4-6: Accumulated seepage using the relative stone content approach for substrate 1 from 1990 to 2016 in Bad Hersfeld ...... 142 Table 4-7: pH values of substrates 1-4 considering the DIN ISO 10390 in 2014 (n=2) and 2016 (n=3) according to different depths ...... 152 Table 4-8: Electrical conductivity (mS/cm) of substrates 1-4 considering the DIN ISO 11265 in 2014 (n=2) and 2016 (n=3) according to different depths ...... 153 Table 4-9: Bulk density (g/cm³) of substrates 1-4 2016 considering different depths (n=6) ...... 154 Table 4-10: Estimated total porosity (%) in substrates 1-4 2016 considering different depths (n=6) ...... 155 Table 4-11: Available and unavailable moisture in the substrates 1-4 (n=6) ...... 156 Table 4-12: Pore size distribution of the substrates 1-4 (n=24) ...... 157 Table 4-13: Saturated hydraulic conductivity in substrates 1-4 2016 considering different depths (n=6)...... 158 Table 4-14: Particle size distribution of substrates 1-4 (n=12) ...... 159

viii

List of tables

Table 4-15: Root-mean-square error (RMSE) and correlation coefficients between observed and forward simulations of seepage and water content using Hydrus-1D from substrates 1-4 ...... 160 Table 4-16: Observed and fitted hydraulic properties of substrate 1 considering five different input data in the inverse solution of Hydrus-1D ...... 160 Table 4-17: Observed and fitted hydraulic properties of substrate2 considering five different inputs in the inverse solution of Hydrus-1D ...... 161 Table 4-18: Observed and fitted hydraulic properties of substrate 3 considering five different inputs in the inverse solution of Hydrus-1D ...... 161 Table 4-19: Observed and fitted hydraulic properties of substrate 4 considering five different inputs in the inverse solution of Hydrus-1D ...... 162 Table 4-20: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly accumulated seepage (104) and one water content measurement (1) ...... 163 Table 4-21: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly accumulated seepage (104) and one water content measurement (1) ...... 164 Table 4-22: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with accumulated seepage by season (4) and one water content measurement (1) ...... 165 Table 4-23: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with accumulated seepage by season (4) and water content measurement (1) ...... 166 Table 4-24: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and retention curve, ψm(Ɵ) (1) ...... 167 Table 4-25: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and retention curve, ψm(Ɵ) (1) ...... 168 Table 4-26: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and seasonal accumulated seepage (4) ...... 169 Table 4-27: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and seasonal accumulated seepage (4) ...... 170

ix

List of tables

Table 4-28: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and monthly accumulated seepage (24) ...... 171 Table 4-29: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and monthly accumulated seepage (24) ...... 172 Table 4-30: Water fluxes for Bad Hersfeld from 1990 to 2016 using hydraulic parameters from substrate 1...... 173 Table 4-31: Components of the simplified water balance equation for Bad Hersfeld from 1990 to 2016 using hydraulic parameters from substrate 1 ...... 174 Table 4-32: Hydraulic properties and accumulated seepage of different substrates using 100 % fine particles 2014 to 2015 in Heringen ...... 175 Table 4-33: Hydraulic properties and accumulated seepage of different substrates using 80 % fine particles 2014 to 2015 in Heringen ...... 175 Table 4-34: Hydraulic properties and accumulated seepage of different substrates using 60 % fine particles 2014 to 2015 in Heringen ...... 175 Table 5-1: Observed hydraulic properties of the substrates 1-4 from 0.0 to 0.64 m depth...... 183 Table 5-2: Total precipitation, effective precipitation, reference evapotranspiration, crop evapotranspiration and water deficit of substrates 1-4 from 2014 to 2016 using CropWat at 10-days interval...... 189 Table 5-3: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 1 from 2014 to 2016 using CropWat under daily water balance ...... 194 Table 5-4: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 2 from 2014 to 2016 using CropWat under daily water balance ...... 194 Table 5-5: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 3 from 2014 to 2016 using CropWat under daily water balance ...... 195 Table 5-6: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 4 from 2014 to 2016 using CropWat under daily water balance ...... 195 Table 5-7: Observed water balance components of substrate 1 during three calendar years at the lysimeter experimental site...... 196 Table 5-8: Observed water balance components of substrate 2 during three calendar years at the lysimeter experimental site...... 196 x

List of tables

Table 5-9: Observed water balance components of substrate 3 during three calendar years at the lysimeter experimental site...... 196 Table 5-10: Observed water balance components of substrate 4 during three calendar years at the lysimeter experimental site...... 197

Table 5-11: Observed and estimated actual evapotranspiration (ETa) and drainage (D) of substrates 1-4 from 2014 to 2016 at the lysimeter experimental site ...... 197 Table 5-12: Observed and estimated water deficit of substrates 1-4 from 2014 to 2016 at the lysimeter experimental site...... 197 Table 5-13: Total precipitation, reference evapotranspiration, crop evapotranspiration and water deficit of substrate 1 for P20, P50 and P80 using CropWat at 10-days interval ...... 201 Table 5-14: Total precipitation, actual evapotranspiration, drainage and water deficit of substrate 1 for P20, P50 and P80 using CropWat for a rainfed field under daily water balance 201 Table 5-15: Total precipitation, reference evapotranspiration, crop evapotranspiration and water

deficit of substrate 1 for P50 and Kc 0.4, 0.6, 0.8, 1.0 using CropWat at 10-days interval ...... 203 Table 5-16: Total precipitation, actual evapotranspiration, drainage, and water deficit of substrate

1 for P50 and Kc 0.4, 0.6, 0.8, 1.0 using CropWat for a rainfed field under daily water balance ...... 204

xi

List of symbols

List of symbols

Symbol Description Unity Equation number q volumetric flow rate m/day (2-4) Q discharge m³/day (2-4) A area perpendicular to the flow m² (2-4) L distance m (2-4)

Ks saturated hydraulic conductivity m/day (2-4) ΔH difference in hydraulic head m (2-4)

Se relative saturation - (2-8) Ɵ volumetric water content cm³/cm³ (2-8)

Ɵr residual water content cm³/cm³ (2-8)

Ɵs water content at soil saturation cm³/cm³ (2-8) h pressure head cm (2-8) α, n, m fitting parameters 1/cm, - (2-8; 2-9) K(h) unsaturated hydraulic conductivity function cm/day (2-9)

Ks saturated hydraulic conductivity cm/day (2-9) l pore-connectivity 0.5 (2-9) ET evapotranspiration mm (2-11) P precipitation mm (2-11) I irrigation mm (2-11) D drainage mm (2-11) ΔW water storage change mm (2-11)

ET0 reference evapotranspiration mm/day (2-57) Rn net radiation at the crop surface MJ/m2/day (2-57) G soil heat flux density MJ/m2/day (2-57) T mean daily air temperature °C (2-57)

u2 wind speed m/s (2-57)

es saturation vapor pressure kPa (2-57)

ea actual vapor pressure kPa (2-57)

es-ea saturation vapor pressure deficit kPa (2-57) Δ slope of the vapor pressure curve kPa/°C (2-57) γ psychrometric constant kPa/°C (2-57)

xii

General introduction

1 General introduction Replacing nutrients in the soil is essential for increasing agricultural productivity, food production and food security (Manning, 2017; Warren, 2016). Organic fertilizers, such as manure and wood ashes have been used for centuries to improve crop growth (Hignett, 1985). Mineral fertilizers have been implemented more recently, mainly after Justus von Liebig (1803-1873) claimed soil crop nutrients could be replaced with mineral sources (van der Ploeg et al., 2005). The essential nutrients plants require are nitrogen, phosphorus and (Ciceri et al., 2015). Nitrogen is synthesized using the Haber–Bosch process, which combines atmospheric nitrogen and hydrogen from fossil fuels under high temperatures and pressure (Manning, 2010). Phosphorus is extracted from marine sediments or igneous rocks (Heckenmüller et al., 2014). Phosphate is produced worldwide and the United States, China, and Morocco are the main producers (al Rawashdeh and Maxwell, 2011). Potassium is mined from underground potash ores or buried evaporites (Rauche, 2015). Potash ores are deposits containing potassium salts soluble in water formed from seawater evaporation (Garrett, 1996; Warren, 2016; Ciceri et al., 2015). The word potash originates from the use of crop ashes to obtain potassium salts (Garrett, 1996). The ashes from forest trees, halophyte plants and marine algae were, up to circa 1860, the main source of potash compounds (Ciceri et al., 2015). In 1861, subsurface potash mining started in Staßfurt, Germany (Warren, 2016; Ciceri et al., 2015). Germany had control of the potash market up to circa 1910, when new potash reserves were found in Spain and in the United States (Ciceri et al., 2015). Potash production started in France in 1910, Spain in 1925, Russia in 1930, the United States in 1931 and Canada in 1960 (Hignett, 1985). Currently the largest potash producers are Canada, Russia and Belarus (Manning, 2017; Ciceri et al., 2015). Germany is the fifth largest potash producer worldwide (Warren, 2016), and the main potash supplier in Europe (Ciceri et al., 2015; Wedig, 2014). The main potassium ore salts found in natural buried evaporites are sylvite, carnallite, kainite and langbeinite (Warren, 2016). Sylvite and carnalite are the most common potash evaporite salts (Warren, 2016). Potassium can also be recovered from surface brines, i.e. the Dead Sea in the Middle East, the Great Salt Lake in the United States and the Salar de Atacama in Chile (Lottermoser, 2010; al Rawashdeh and Maxwell, 2014). Most potash salts are used as fertilizers, 90-95 %. (Warren, 2016; al Rawashdeh et al., 2016). Potash is also found in chemicals, glasses, ceramics, detergents, soups and synthetic rubber (Warren, 2016; Ciceri et al., 2015). Presently, circa 150 countries use potash fertilizers in agriculture, the United States being the main consumer (Warren, 2016). China, Brazil and India are the main potash importers (Warren, 2016). Three main methods are used to extract potassium from underground ore deposits. Conventional shaft mining is performed during 80 % of potassium extraction at a depth of up to 1,100 m. Solution 1

General introduction mining comprises 6% of potassium extraction and is implemented at depths higher than 1500 m. The remaining potash extraction is performed using solar evaporation from natural brines (IPI, 2014; Lottermoser, 2010; al Rawashdeh and Maxwell, 2014). Potash mining produces solid and liquid wastes with a high concentration of sodium chloride (Warren, 2016; Reid et al., 2004; Thorpe et al, 1991; Turk, 1970). According to Podlacha (1999), potash mining residues represent circa 75 % of the total mined volume. The liquids are pressed underground or injected in surface waters to be transported to oceans, whereas the solids are heaped near processing facilities or backfilled in mining voids (Rauche, 2015). When piled aboveground, precipitation dissolves the salts and generates brines that are collected and injected into deep wells or rivers (European Commission, 2009). In Germany, circa 500 ha are covered with potash tailings piles (Rauche, 2015). Several studies have examined the environmental impacts of brine injection from potash mining in water systems. These studies mainly estimate the fauna and flora of rivers (Coring and Bäthe, 2011; Bäthe and Coring, 2011) and the chemical composition of the water courses surrounding the potash mining facilities (Braukmann and Böhme, 2011; Cañedo-Argüelles et al. 2013; Cañedo- Argüelles et al., 2017; Ladrera et al., 2017). Additional studies have evaluated the technical feasibility of using technogenic substrates as evapotranspiration covers on potash tailings piles. An evapotranspiration cover aims to store the precipitation water in a soil or soil substitute reservoir and then transport the moisture back to the atmosphere using perennial crops (Hauser, 2009). Technogenic substrates refer to materials created by humans or from human activities such as wastes, slags and ashes (Blume et al., 2016). Podlacha (1999) assessed a mixture of soil and ashes; Scheer (2001), Hermsmeyer (2001) and Nessing (2005) investigated the use of aluminium recycling products blended with ashes. The aluminum recycling by-products need circa 1361 mm of rain to leache out part of the salt content and proceed with seeding (Hermsmeyer, 2001). In contrast, the use of ashes increases the consolidation and stability of the substrates, due to their pozzolanic properties (Podlacha, 1999). Pozzolanic properties are characterized by the ability of fine and dispersed silicates and aluminosilicate substances to react with calcium hydroxide in the presence of water and form solid and stable compounds, such as calcium silicate hydrates or calcium aluminate hydrates (Navrátilová and Rovnaníková, 2016). Several industrial by-products have pozzolanic properties, such as fly ashes from coal combustion and ashes from biomass combustion (Navrátilová and Rovnaníková, 2016). During coal combustion, fly ashes comprise 62 % of the residues, followed by flue gas desulfurization from air pollution control (19 %), and boiler slag and bottom ashes (18 %) (Vinai et al., 2013; National Research Council, 2006). In the European Union, circa 100 million tons of coal combustion residues are produced annually (Spliethoff, 2010). These residues are used in building construction, roads, and applied as filler in mining voids (Feuerborn, 2011). 2

General introduction

Since June 2005 municipal wastes can no longer be landfilled without pre-treatment in Germany (Nelles et al., 2016). Currently, the household waste, not recycled or composted, must be incinerated. This thermal treatment of municipal solid wastes generates residues which are classified as bottom ashes, fly ashes and boiler ashes (Inkaew et al., 2016). Fly ashes are removed prior to the air pollution control structures, boiler ashes are captured from heat recovery systems and bottom ashes are discharged from the burning grate of the incinerator (Chandler et al., 1997). Bottom ash is the main product, amounting to 250 kg for every ton of solid incineration waste (Holm and Simon, 2017; Inkaew et al., 2016). Circa 5 million tons of household incineration bottom ashes are produced annually in Germany, which are used for construction or landfilled (Holm and Simon, 2017). Using the data collected from these past experiments, this study seeks to further the research on evapotranspiration covers for potash tailings piles using technogenic substrates made of household waste incineration slags and coal combustion residues. Specifically, this thesis focuses on the results from an experiment conducted in Heringen, Germany where 8 non-weighing lysimeters were installed in July 2013 and inspected through to 2016. The experimental site included a weather station, 4 rain gauges installed on the soil surface and 5 rain gauges installed at 1 m high to monitor weather conditions.

1.1 Objectives and hypotheses of the research The present study had three main objectives, which were distributed over three chapters. The first main objective was (i) to assess the evapotranspiration and drainage of the four technogenic substrates made of household waste incineration slags and coal combustion residues during 2014 and 2015. These substrates were placed in eight non-weighable lysimeters installed on a potash tailings pile located in Heringen (Werra). Furthermore, the weather parameters of the experimental site were studied, such as precipitation, wind speed, air humidity, solar radiation, air and soil temperature. Moreover, the reference evapotranspiration and crop evapotranspiration using the FAO standard method were determined and compared with Haude´s potential evapotranspiration. The potash tailings covers were expected to increase evapotranspiration and decrease the drainage of potash tailing piles. Differences in the water fluxes among the different substrates were also predicted. The second study (ii) aimed to simulate evapotranspiration and drainage of the evapotranspiration covers using Hydrus-1D. For this, the physical-hydraulic properties, pH and electrical conductivity of the technogenic substrates were investigated. Then, direct simulations of water fluxes were conducted and the van Genuchten hydraulic parameters were optimized using seepage, water content and water retention curve measurements in the inverse solution of Hydrus- 1D model. After, the calibrated model was validated using historical daily weather parameters from 3

General introduction a neighbouring weather station 20 km away. Then the seepage and root water uptake of the evapotranspiration covers were simulated using different ratios of fine fraction, soil textures and crop parameters. We hyphotetized that an increase in the rate of fine fraction in the substrates will likely decrease the drainage rate of the evapotranspiration covers. The third and last study (iii) investigated the evapotranspiration deficit of potash tailings piles’ covers using the CropWat model. This study determined the effective precipitation, crop evapotranspiration, actual evapotranspiration and water deficits of the substrates from 2014 to 2016. In addition, 30 years of historical weather data were used to estimate the water deficit for different precipitation regimes and crop coefficients. We expected that evapotranspiration deficits occur in periods of high potential evapotranspiration, such as those found in spring and summer months.

1.2 Thesis structure The first chapter provides a general introduction about the issues regarding potash mining and its environmental concerns such as brines and salt tailings. Chapter two presents the state of the art comprising the main theories on evapotranspiration, drainage, evapotranspiration covers, minining activities and technogenic substrates. The third chapter describes the water balance assessment of eight non-weighable lysimeters, considering four different technogenic substrates, during 2014 and 2015. This study was published in the Journal of Environmental Management, volume 196, pages 633-643. The fourth chapter describes the simulations of evapotranspiration and drainage of the evapotranspiration covers using Hydrus-1D. Continuing, the fifth chapter elaborates on the water deficits of the evapotranspiration covers using the CropWat model. The sixth chapter reviews the main findings of the research. This section is followed by the final conclusions and suggestions for future studies. Part of the fourth and the fifth chapters will be submitted to scientific journals in the coming weeks.

1.3 References al Rawashdeh, R., Maxwell, P., 2011. The evolution and prospects of the phosphate industry. Miner. Econ. 24, 15-27. Doi: 10.1007/s13563-011-0003-8. al Rawashdeh, R., Maxwell, P., 2014. Analyzing the world potash industry. Res. Policy 41, 143- 151. Doi: 10.1016/j.resourpol.2014.05.004. al Rawashdeh, R., Xavier-Oliveira, E., Maxwell, P., 2016. The potash market and its future prospects. Res. Policy 47, 154-163. Doi: 10.1016/j.resourpol.2016.01.011. Bäthe, J., Coring, E., 2011. Biological effects of anthropogenic salt-load on the aquatic fauna: A synthesis of 17 years of biological survey on the rivers Werra and Weser. Limnologica 41, 125- 133. Doi: 10.1016/j.limno.2010.07.005. 4

General introduction

Blume, H.P., Brümmer, G.W., Fleige, H., Horn, R., Kandeler, E., Kögel-Knabner, I., et al., 2016. Scheffer/Schachtschabel Soil Science. (1 ed.). Berlin: Springer. Braukmann, U., Böhme, D., 2011. Salt pollution of the middle and lower sections of the river Werra (Germany) and its impact on benthic macroinvertebrates. Limnologica 41, 113-124. Doi: 10.1016/j.limno.2010.09.003. Cañedo-Argüelles, M., Brucet, S., Carrasco, S., Flor-Arnau, N., Ordeix, M., Ponsá, S., Coring, E., 2017. Effects of potash mining on river ecosystems: An experimental study. Environ. Pollut. 224, 759-770. Doi: 10.1016/j.envpol.2016.12.072. Cañedo-Argüelles, M., Kefford, B.J., Piscart, C., Prat, N., Schäfer, R.B., Schulz, C., 2013. Salinisation of rivers: An urgent ecological issue. Review. Environ. Pollut. 173, 157-167. Doi: 10.1016/j.envpol.2012.10.011. Chandler, A.J., Eighmy, T.T., Hjelmar, O., Kosson, D.S., Sawell, S.E., Vehlow, J., Sloot, H.A., Hartlén, J., 1997. Municipal solid waste incinerator residues. Amsterdam: Elsevier. Ciceri, D., Manning, D.A.C., Allanore, A., 2015. Historical and technical developments of potassium resources. Sci. Total Environ. 502, 590-601. Doi: 10.1016/j.scitotenv.2014.09.013. Coring, E., Bäthe, J., 2011. Effects of reduced salt concentrations on plant communities in the River Werra (Germany). Limnologica 41, 134-142. Doi: 10.1016/j.limno.2010.08.004. European Commission, 2009. Reference document on best available techniques for management of tailings and waste-rock in mining activities. http://eippcb.jrc.ec.europa.eu/reference/BREF/mmr_adopted_0109.pdf (accessed 17 August 2016). Feuerborn, H., 2011. Coal combustion products in Europe - An update on production and utilization, standardization and regulation. World of Coal Ash Conference (WOCA), May 9-12, 2011, in Denver, Colorado, USA. http://www.flyash.info/2011/007-feuerborn-2011.pdf (accessed 02 September 2016). Garrett, D.E., 1996. Potash. Deposits, processing, properties and uses. Dordrecht: Springer. Hauser, V.L., 2009. Evapotranspiration covers for landfills and waste sites. Boca Raton: CRC Press. Heckenmüller, M., Narita, D., Klepper, G., 2014. Global availability of phosphorus and its implications for global food supply. An economic overview. Kiel Working Paper, 1897 (accessed 12 December 2017). Hermsmeyer, D., 2001. Soil physical and hydrological evaluation of aluminum recycling by- product as an infiltration barrier for potash mine tailings (Doctoral Dissertation). Hanover University, Welfengarten. Hignett T.P., 1985. History of chemical fertilizers. In Hignett T.P., 1985. Fertilizer Manual. Developments in Plant and Soil Sciences, vol 15. Dordrecht: Springer.

5

General introduction

Holm, O., Simon, F., 2017. Innovative treatment trains of bottom ash (BA) from municipal solid waste incineration (MSWI) in Germany. Waste Manage. 59, 229-236. Doi: 10.1016/j.wasman.2016.09.004. Inkaew, K., Saffarzadeh, A., Shimaoka, T., 2016. Modeling the formation of the quench product in municipal solid waste incineration (MSWI) bottom ash. Waste manage. 52, 159-168. Doi: 10.1016/j.wasman.2016.03.019. International Potash Institute (IPI), 2014. Production and use of potassium chloride. http://www.ipipotash.org/udocs/Chap-1_potash_production.pdf (accessed 30 March 2014). Ladrera, R., Cañedo-Argüelles, M., Prat, N., 2017. Impact of potash mining in streams. The Llobregat basin (northeast Spain) as a case study. J. Limnol. Doi: 10.4081/jlimnol.2016.1525. Lottermoser, B., 2010. Mine wastes: Characterization, treatment and environmental impacts. Berlin: Springer. Manning, D.A.C., 2017. Innovation in resourcing geological materials as crop nutrients. Nat. Resour. Res. 24, 1-11. Doi: 10.1007/s11053-017-9347-2. Manning, D.A.C., 2010. Mineral sources of potassium for plant nutrition. A review. Agron. Sustainable Dev. 30, 281-294. Doi: 10.1051/agro/2009023. National Research Council (U.S.), 2006. Managing coal combustion residues in mines. Washington: National Academies Press. Navrátilová, E., Rovnaníková, P., 2016. Pozzolanic properties of brick powders and their effect on the properties of modified lime mortars. Constr. Build. Mater. 120, 530-539. Doi: 10.1016/j.conbuildmat.2016.05.062. Nelles, M., Grünes, J., Morscheck, G., 2016. Waste management in Germany - Development to a sustainable circular economy? Procedia Environ. Sci. 35, 6-14. Doi: 10.1016/j.proenv.2016.07.001. Niessing, S., 2005. Begrünungsmaßnahmen auf der Rückstandshalde des Kaliwerkes - Sigmundshall in Bokeloh (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 25/2005, Universität Kassel, Witzenhausen. Podlacha, G., 1999. Untersuchungen zur Substratandeckung mit geringen Schichtstärken aus Bodenaushub-Wirbelschichtasche-Gemischen und ihrer Begrünung (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 16/1999, Universität Kassel, Witzenhausen. Rauche, H., 2015. Die Kaliindustrie im 21. Jahrhundert. Stand der Technik bei der Rohstoffgewinnung und der Rohstoffaufbereitung sowie bei der Entsorgung der dabei anfallenden Rückstände. (1. Aufl.). Berlin: Springer. Reid, K.W., Getzlaf, M.N., 2004. Decommissioning planning for Saskatchewan's potash mines. Conference Paper: British Columbia Mine Reclamation Symposium. University of British Columbia, Vancouver, Canada. Doi: 10.14288/1.0042463. 6

General introduction

Scheer, T., 2001. Rekultivierung von Rueckstandshalden der Kaliindustrie. Untersuchungen zur Nutzbarkeit aufbereiteter Salzschlacke der Sekundaeraluminium-Industrie als Rekultivierungsmaterial einer Kali-Rückstandshalde (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 20/2001, Universität Kassel, Witzenhausen. Spliethoff, H., 2010. Power generation from solid fuels. Berlin: Springer. Thorpe, M.B., Neal, D., 1991. Revegetation of saline land caused by potash mining activity. Conference Paper: British Columbia Mine Reclamation Symposium. University of British Columbia, Vancouver, Canada. Doi: 10.14288/1.0042169. Turk, L.J., 1970. Evaporation of brine: A field study on the Bonneville Salt Flats, Utah. Water Resour. Res. 6, 1209-1215. Doi: 10.1029/WR006i004p01209. van der Ploeg, R.R., Böhm, W., Kirkham, M.B., 2005. Liebig, Justus Von. In: Hillel, D., 2005. Encyclopedia of soils in the environment. Amsterdam: Elsevier, p. 343-349. Vinai, R., Lawane, A., Minane, J.R., Amadou, A., 2013. Coal combustion residues valorization. Research and development on compressed brick production. Constr. Build. Mater. 40, 1088- 1096. Doi: 10.1016/j.conbuildmat.2012.11.096. Warren, J.K., 2016. Evaporites. A geological compendium. (2 ed.). Heidelberg: Springer. Wedig, M., 2014. German mining industry overview. Min. Rep. 150, 90-93. Doi: 10.1002/mire.201400008.

7

State of the art

2 State of the art The state of the art contains a literature review on evapotranspiration, drainage, evapotranspiration covers and technogenic substrates. This section will also discuss mining and potash tailings in the Werra Region where the lysimeter experiment was located.

2.1 Water cycle The water cycle is essential for life on earth, connecting water reservoirs in the lithosphere, atmosphere and oceans through solar radiation (Harding et al 2014; Marshall, 2014). The water cycle also helps purify water and transports sediments (Hendriks, 2010). Besides this, it promotes ecosystem services, energy generation, agricultural and industrial production and maintains human health (Harding et al 2014; Narasimhan, 2009; Rast et al., 2014; Hendriks, 2010). Overall the water cycle starts in the oceans, where water evaporates and moisture vapor condenses in the atmosphere and precipitates (Hendriks, 2010; Seiler and Gat, 2007). However, not all precipitation falls in the oceans (Seiler and Gat, 2007). The total evaporation in the oceans is circa 1393 mm/year and precipitation 1269 mm/year (Seiler and Gat, 2007). Some atmospheric moisture from the oceans, circa 9%, moves with the wind and precipitates over exorheic regions (Seiler and Gat, 2007). After, the water moves back to the oceans through surface and subsurface runoff where the water cycle reinitiates (Hendriks, 2010; Seiler and Gat, 2007). Further water circulation is observed on the continents, where moisture evapotranspirates and precipitates without reaching the oceans (Seiler and Gat, 2007). This closed water circulation system is known as endorheic (Seiler and Gat, 2007). Endorheic regions are located at lakes with high salt concentrations such as the Dead Sea in The Middle-East and the Great Salt lakes in the United States (Bajraktari et al., 2017). Water cycle studies provide information about the spatial and temporal variations of hydrological cycle components (Rast et. al., 2014; Harding et al 2014; Güntner et al., 2007). Studies performed on a local scale aim to assess the water consumption of a few square meters, such as a lysimeter or experimental plot (Hopmans and Schoups, 2006). Studies conducted at a regional scale provide information about hydrological components in agricultural fields and can estimate river levels (Narasimhan, 2009; Hopmans and Schoups, 2006). In contrast, global scale studies estimate the rate of fresh water availability (Seiler and Gat, 2007). Some of the main processes involved in the water cycle are precipitation, evapotranspiration, runoff, infiltration and soil moisture redistribution (Hendriks, 2010).

2.1.1 Precipitation Precipitation is the main component of the water cycle (Tukimat et al., 2012). Precipitation occurs when water droplets in the atmosphere overcome air resistance and fall on the earth’s surface 8

State of the art

(Hendriks, 2010). Precipitation can be in the form of rain (liquid) or snow (solid) (Shuttleworth, 2012). Dew, fog drips and frost are also considered precipitation (Shuttleworth, 2012). Moreover, amount, frequency and intensity are precipitation’s main parameters (Seiler and Gat, 2007). Precipitation intensity can be classified as weak (< 1.5 mm/h), medium (2.6-7.5 mm/h), and strong (> 7.6 mm/h) and varies depending on the region (Seiler and Gat, 2007). Higher variability is found in the tropics, arid and semi-arid regions, whereas in cold and temperate climates, rain frequency is more constant (Seiler and Gat, 2007). Precipitation amount is distributed unevenly around the globe due to the variations of solar radiation, wind movement, albedo and topography (Seiler and Gat, 2007). Precipitation is high under temperate and tropical regions whereas warm and cold areas have lower precipitation levels (Seiler and Gat, 2007). The highest mean precipitation is found in South America, 1595 mm/year, and the lowest in Africa and Asia, circa 740 mm/year (Shuttleworth, 2012). Globally, mean annual precipitation ranges from 1000 to 1123 mm (Kidd and Huffman, 2011), circa 1,269 mm is measured in the oceans, 924 mm on the exorheic continental areas and 300 mm in endorheic inland regions (Seiler and Gat, 2007). Precipitation can be measured using storage containers or gauges, such as tipping bucket gauges, weighing gauges or acoustic rain gauges (Kidd and Huffman, 2011; Shuttleworth, 2012). Further estimations of precipitation at a global level can be performed using satellites (Kidd and Huffman, 2011).

2.1.2 Infiltration Infiltration is a process in which water crosses through the soil surface during precipitation (Brady and Weil, 2014). Infiltration is one of the main hydrological processes because it determines the portion of rain that will flow over the surface and the water amount available for crops (Hillel, 1998). Infiltrated water can also contribute to ground water recharge (Hopmans, 2010). At field conditions, it is possible to evaluate both the cumulative infiltration and the infiltration rate (Blume et al., 2016). Cumulative infiltration represents the potential water volume that can infiltrate an area over a specific time (Blume et al., 2016) and the infiltration rate represents the rate at which water enters the soil surface (Bume et al., 2016). Measuring the infiltration rate is done using the following expression (Brady and Weil, 2014): 푄 (2-1) 𝑖 = 퐴 . 푡 Where i is the infiltration rate (m/s), Q is the infiltration volume (m³), A is the area (m²) and t is the time (s). The infiltration rate decreases with time due to higher levels of moisture in the soil profile (Lal and Shukla, 2004; Blume et al., 2016). The maximum infiltration rate of the soil or substrate matrix is obtained at initial infiltration time (Hopmans, 2010) and decreases when water flow advances to a steady flux. In this period (steady flux), the soils have reached saturated hydraulic conductivity 9

State of the art

(Lal and Shukla, 2004), which represents the capability of the soil to transfer water (Hopmans, 2010). Generally, the moisture distribution in the soil profile can be divided into four zones, saturation zone, transition zone, transport zone and wetting front (Blume et al., 2016). The saturation and the transition zone make up a few centimeters (Blume et al., 2016). However, the transport zone’s length increases with time (Blume et al., 2016). At field conditions, if the precipitation is higher than the maximum instantaneous infiltration capacity, there is ponding on the surface or runoff and the infiltration is equal to the infiltration capacity of the soil (profile controlled) (Lal and Shukla, 2004). If precipitation is lower than the instantaneous infiltration capacity, then the infiltration rate is regulated by rain intensity (flux controlled) (Lal and Shukla, 2004). Infiltration capacity is the maximum water infiltration that soils can absorb under ponded water (Hopmans, 2010). Infiltration can be measured or estimated. Common measurements include the use of a double ring infiltrometer (Brad and Weil, 2014), disc infiltrometer and a minidisk infiltrometer (Kirkham, 2014). A double ring infiltrometer has an outer ring, circa 30-50 cm in diameter and an inner ring, circa 20-30 cm in diameter (Hillel, 1998; Lal and Shukla, 2004). The metal rings are pressed into the soil and the water depth is ponded (Zhang et al., 2017). Then the volume of water infiltrated over a period is measured (Hillel, 1998). The disc and the minidisk infiltrometer measure the infiltration or hydraulic conductivity under water tension (Kirkham, 2014; Zhang et al., 2017). The disc infiltrometer minimizes the interference of preferential flow on the infiltration measurements because the water is under tension (Kirkham, 2014). Preferential flow is characterized by high speed water fluxes due to cracks, decay of roots or shrinkage (Horton et al., 2016). These secondary pores increase the water velocity in comparison with the soil matrix (Horton et al., 2016). Downward infiltration can also be estimated using physical or empirical expressions. Physical models are based on soil properties, which are difficult to measure and present a high variability on the soil surface, such as the Philip and the Green and Ampt equations (Blume et. al., 2016; Kirkham, 2014). The commonly used empirical or statistical expressions are the Kostiakov equation and the Horton equation (Kirkham, 2014). The infiltration rate varies according to the initial moisture, texture, structure, rain intensity, rain amount, soil cracks (Hillel, 1998), vegetation cover, hydraulic conductivity, hydraulic gradient (Lal and Shukla, 2004), surface slope (Morbidelli et al., 2015) and soil layers (Hillel, 1998). Another factor that interferes with infiltration is crust formation or sealing on the surface (Hopmans, 2010). Crust or surface sealing are caused by the exposure of the soil aggregates to perturbations from agricultural practices, i.e., tillage or traffic, or weather conditions, such as precipitation (Lal and Shukla, 2004). Crust and sealing can also be classified into chemical, biological and physical crusts (Lal and Shukla, 2004). Chemical crusts are a product of salt accumulation on the surface, generally 10

State of the art seen in arid regions (Lal and Shukla, 2004). Biological crusts are due to the growth of algae on the surface (Lal and Shukla, 2004). And physical crusts are due to rain drops’ impacts or mineral deposition from fine particles (Lal and Shukla, 2004). These disturbances may disintegrate the aggregates and clog the surface pores decreasing the infiltration capacity (Lal and Shukla, 2004). With regards to texture, coarse soils may have a higher infiltration rate due to larger pores (Lal and Shukla, 2004). However, cracks in contact with the atmosphere (Zhang et al., 2017) in clay soils can also speed up the infiltration rate (Blume et al., 2016). The vegetation cover and organic matter on the surface also help minimize the impact of rain drops and avoid surface sealing (Hopmans, 2010).

2.1.3 Redistribution After infiltration, the water from precipitation or irrigation continues to move within the soil profile through a process called redistribution (Hillel, 1998; Kirkham, 2014). In this process, the soil surface gets drier and deeper layers receive moisture (Lal and Shukla, 2004). Redistribution determines the water volume retained at a determined depth and the water volume that flows to deeper profiles (Hillel, 1998). If the soil profile is wetted up to saturation, the water movement after infiltration is called drainage (Hillel, 1998). In unsaturated profiles, Seiler and Gat (2007) term percolation the water fluxes between the surface and the ground water since it is more associated with capillary than gravity gradients. Bethune et al. (2008) define deep percolation the water flow below the root zone. This moisture represents the field capacity excess from the upper soil layers (Bethune et al., 2008).

2.1.4 Runoff Runoff contributes to water discharge into rivers due to overland and subsurface flow. Subsurface flow is subdivided into base flow (ground-water flow) and inter-flow (Seiler and Gat, 2007). Surface overflow and inter-flow consist of rapid water fluxes (meters per second or hour) albeit ground-water flow is a slow process (meters per day or year) because water interacts with mineral surfaces (Seiler and Gat, 2007). Therefore, studies on ground water recharge should consider time series longer than 25 years (Seiler and Gat, 2007). Hillel (1998) classifies the surface flow as overland and stream flow (river flow). Water flow over the soil surface occurs when the infiltration capacity of the soils is lower than the application rate of water (Hillel, 1998). In this case, rain water may pond or overflow onto the surface (Hillel, 1998). Higher surface overflow is observed in sloped areas (Hillel, 1998). Surface runoff increases erosion and the transportation of organic matter, sediments, soluble nutrients, chemicals and microorganisms (Wang et al., 2017; Merten et al., 2015; Bradford et al., 2015). These sediments and pollutants can reduce the quality of surface water courses (Rose, 2004). Therefore, cover crops 11

State of the art and mulching are used to increase the water infiltration and reduce raindrop impact on the soil surface (Hillel, 1998). Crops improve surface roughness and soil structure (Zhang et al., 2015). Additionally, crop roots at different densities and depths improve the infiltration time and reduce the leaching of nutrients (Wang et al., 2017; Zhang et al., 2015). The use of crops to reduce the negative impact of runoff has been implemented in agricultural areas (Merten et al., 2015), mining landscapes (Zhang et al., 2015) and urban regions on vegetated roofs (Wang et al., 2017). Yet surface runoff also contributes to the discharge of rivers (Hillel, 1998; Hendriks, 2010), which is an important part of the water cycle. Measuring surface runoff can be done using runoff plots (Rose, 2004). These runoff plots are circa 500 m² and collect runoff water in tipping bucket systems (Rose, 2004). Large water flow measurements can also be performed using weirs and flumes (Rose, 2004). Also, mathematical models such as Hydrus have been used to estimate the surface runoff (Simunek et al., 2013, Caiqiong and Jun, 2016; Bradford et al., 2015). At the watershed level, i.e., a land area which drains the incoming precipitation to the same place (Edwards et al., 2015), the Curve Number method from the US Department of Agriculture (USDA, 1986) is commonly used to evaluate the surface runoff (Satheeshkumar et al., 2017; Mishra et al., 2014). This method consists of using an empirical equation which predicts the runoff from precipitation and an S factor that accounts for soil water content, vegetation cover, and land use (Mishra et al., 2014). Inter-flow takes place in opposing hydraulic conductivities at unsaturated zones which transforms vertical into horizontal flow (Seiler and Gat, 2007). Ground-water flow occurs in a saturated zone below the surface and responds to gravitational gradients (Seiler and Gat, 2007). Ground-water recharge considers the residual water flow from surface runoff, interflow and evapotranspiration (Seiler and Gat, 2007). In temperate climates, the ground water recharge represents circa 10 % of the precipitation and in arid regions this rate is circa 5 % (Seiler and Gat, 2007). Ground water recharge can be estimated using mass balance methods (Seiler and Gat, 2007) and the ground-water level is measured using piezometers (Hillel, 1998). Ground water recharge is an important issue due to increasing population and the consequent pressure on water resources for living and food production (Seiler and Gat, 2007). Presently, circa 40 % of human’s water consumption originates from ground water (Seiler and Gat, 2007). Likewise, around 38 % of the irrigated area is based on ground water withdrawal (Siebert et al., 2010).

2.1.5 Evapotranspiration Evapotranspiration is the second most important component in the water cycle (Tukimat et al., 2012). On continents, evapotranspiration represents circa 60.6 % of precipitation (Narasimhan, 2009) whereas, globally, evapotranspiration accounts for all precipitation volume (Seiler and Gat, 2007). Runoff accounts for 39.4 % of continental precipitation (Narasimhan, 2009). 12

State of the art

Evapotranspiration comprises the transpiration of water from the crops and the evaporation of water from the soil (Allen et al., 1998).

2.1.5.1 Transpiration Transpiration represents the water loss by crop leaves through stomata (Taiz et al., 2015). Transpiration occurs via the steady movement of water from the soil to the atmosphere through crops as an effect of the water potential gradient (Kirkham, 2014). Water potential gradient refers to the differences in water status from one point to another within the soil-plant-atmosphere continuum (Hopmans, 2010). The direction of water is from a higher to a lower potential (more negative) and the water flux volume is proportional to the magnitude of the potential gradient and crop resistances (Taiz et al., 2015). Larcher (2003) illustrates the water potential gradient in the soil-plant-atmosphere continuum for a dry and humid atmosphere. For a humid atmosphere, the water potential in the soil could reach 0 MPa, in the roots -0.6 MPa, in the leaves -1 MPa and in the atmosphere, with a relative air humidity of 93 % -10 MPa (Larcher, 2003). However, for dry air (48 % air humidity) the air water potential approaches -100 MPa (Larcher, 2003). Therefore, the highest water potential gradient is found between the leaves and the atmosphere (Kirkham, 2014). The water potential on the soil- plant-atmosphere continuum varies according to the time of the day (Kirkham, 2014), weather conditions (Blume et al., 2016), vegetation and available moisture (Taiz et al., 2015). Well-watered crops show higher water potential (less negative) than water-stressed crops (Taiz et al., 2015). Maximum crop water potential is found in the early morning when plants rehydrate during the night and the lowest is found in the early afternoon (Kirkham, 2014).

2.1.5.2 Evaporation Evaporation of water from the soil is the process in which water changes to vapor (Lal and Shukla, 2004). On the soil surface the evaporation follows an initial, intermediate and final stage (Lal and Shukla, 2004). During the initial stage or just after precipitation, when the bare soil or substrate has enough moisture and a high level of hydraulic conductivity, evaporation is high and is dominated by weather conditions, such as temperature, relative air humidity and wind (Lal and Shukla, 2004). When the hydraulic conductivity decreases, it is compensated by an increase in hydraulic gradient (Lal and Shukla, 2004). At this stage, evaporation is at the potential rate and is constant. In the intermediate stage evaporation starts to decrease because water cannot evaporate at the potential rate and the hydraulic gradient does not compensate for the reduction in hydraulic conductivity (Lal and Shukla, 2004). In the final stage, liquid-water transportation is replaced by vapor diffusion (Lal and Shukla, 2004).

13

State of the art

The evaporation rate is associated with the rain or irrigation amount and frequency, pore connectivity (Seiler and Gat, 2007), soil water availability, air temperature, wind, solar radiation and degree of shading (Allen et al., 1998). Warmer temperatures, frequent rains and wind speed lead to higher evaporation rates because these conditions provide the energy to evaporate the liquid water and increase the vapor pressure deficit between the soil surface and the atmosphere (Seiler and Gat, 2007). Allen et al. (1998) highlight that up to 50 % of the moisture at the permanent wilting point can be evaporated. In addition, a depth of 15 cm can be considered during evaporation processes (Allen et al., 1998). In field crops evaporation is higher at the initial development of the crops and lower according to ground cover (Allen et al., 1998). For instance, during the initial growth circa 10 % of the surface is covered by the crops (Allen et al., 1998). Whereas before flowering the soil is almost complete shaded by the crops (Allen et al., 1998). Evaporation also occurs on the crop leaves before the rain reaches the soil. The precipitation water that evaporates from the leaves is known as interception loss (Edwards et al., 2015). Interception is low in grasses and annual crops and higher in forests because of a more expansive canopy and a greater leaf area and leaf density, facilitating precipitation retention (Edwards et al., 2015). The interception varies according to the rain intensity and weather conditions (Edwards et al., 2015). Interception in forests may range from 10 to 50 % of precipitation volume (Larcher, 2003). The precipitation intercepted by grasslands varies from 3-5 % and for cultivated fields it is less than 10 % (Larcher, 2003). According to Larcher (2003) most of the precipitation water retained by trees is lost to evaporation (Larcher, 2003). To avoid evaporation, Buckingham (1907) recommends the use of mulch. Mulch has a very low hydraulic conductivity, therefore minimizing the capillary action and evaporation from the surface (Buckingham, 1907). In soils that dry rapidly (arid climate), a moisture barrier is formed on the surface (Buckingham, 1907). This occurs because the hydraulic conductivity of the dry soil is very low (Buckingham, 1907). In dry soils the hydraulic conductivity is characterized by thin water films over the soil grains (Buckingham, 1907). This film can break and interrupt the water flow (Buckingham, 1907).

2.1.5.3 Water movement on the soil-plant-atmosphere continuum

Water movement on the soil On the soil, water flow occurs in saturated and unsaturated flow domains (Radcliffe and Simunek, 2010). Under saturated conditions the pores are filled with water (Radcliffe and Simunek, 2010). Whereas under unsaturated conditions, air, water and solids interact in the soil matrix (Lal and Shukla, 2004).

14

State of the art

Two main expressions are used to explain the water flow in saturated conditions: Hagen– Poiseuille and Darcy´s equation (Radcliffe and Simunek, 2010). The Hagen–Poiseuille expression is used to study the water flow in single tubes, with a uniform radius and laminar flow (Radcliffe and Simunek, 2010). In contrast, the Darcy equation estimates the water flow in the entire saturated flow domain, which consists of different pore diameters. Additionally, Darcy´s equation assumes laminar and steady water flow (Hendriks, 2010). Laminar fluxes are found when water flows in sheets with uniform velocity (Shukla, 2014) and turbulent fluxes are found when water flows radially or axially (Lal and Shukla, 2004). Laminar and turbulent fluxes are estimated using Reynolds number. Reynolds number represents the ratio between inertia forces to viscous forces, obtained through the following expression (Horton et al., 2016): 퐹표푟푐푒푠 표푓 𝑖푛푒푟푡𝑖푎 휌 . 휈 . 푟 . (2-2) 푅푒 = = 퐹표푟푐푒푠 표푓 푣𝑖푠푐표푠𝑖푡푦 휂 Where Re is the Reynolds number (dimensionless), ρ is the density of the fluid, v is the mean flow velocity, η is the dynamic viscosity, and r represents the radius of the pore channel (Lal and Shukla, 2004). In tubes the laminar flow is considered to have a Reynolds’ number of ≤ 2000 and a turbulent flow has a Reynolds’ number larger than 4000. Reynolds’ numbers between 2000 and 4000 are for transitional flow regimes (Lal and Shukla, 2004). Clay, silt and sandy soils have Reynolds number lower than 1, i.e., laminar flow, whereas turbulent flow may be verified in gravels (Horton et al., 2016). The low Reynold’s number in soils is due to irregular diameters of the pores (Lal and Shukla, 2004). Darcy’s equation, published in 1856, considers the saturated water flux associated with the coefficient of proportionality, i.e. saturated hydraulic conductivity, and hydraulic gradient. Hagen–Poiseuille equation ∆푝 (2-3) 푞 = 휋 . 푅4 . 8 . 휂 . 퐿 Where q is the volumetric flow rate, R is the radius of the cylinder, 휂 is the coefficient of dynamic water viscosity, Δp is the difference in pressure between two points along the cylinder separated by a distance L (Horton et al., 2016).

Darcy equation 푄 ∆퐻 (2-4) 푞 = = − 퐾 . 퐴 푠 퐿 Where q is the volumetric flow rate (m/day), Q is the volume flux (m³/day), A is the area (m²),

L is the distance in which the hydraulic gradient occurs (m), Ks is the saturated hydraulic conductivity (m/day), ΔH is the difference in hydraulic head at the water receiving end (h2) minus

15

State of the art

the hydraulic head at water dispatching (h1) (Hendriks, 2010). The negative signal is to show that the flux is positive in the direction of the water flow (Radcliffe and Simunek, 2010). The hydraulic head (H) integrates gravity and the pressure head (m). The water movement in unsaturated conditions is explained by the Darcy-Buckingham equation, the continuity equation and Richard’s equation (Hendriks, 2010). Darcy-Buckingham improved the Darcy equation by considering the unsaturated hydraulic conductivity for unsaturated flow (Shukla, 2014). The continuity equation is also known for being the water balance equation, accounting for inputs, outputs and water storage change in a system (Hendriks, 2010). Richard’s equation has been widely used because it combines the Darcy-Buckingham and the continuity equation and explains the transient water flow in unsaturated soils (Hendriks, 2010; Radcliffe and Simunek, 2010).

Darcy-Buckingham equation 휕퐻 휕ℎ 휕ℎ (2-5) 푞 = −퐾 = −퐾 −퐾 = −퐾 . ( + 1) (ℎ) 휕푧 (ℎ) 휕푧 (ℎ) (ℎ) 휕푧

Where q is the flux density (cm/day), K(h) is the unsaturated hydraulic conductivity (cm/day), h is the negative pressure head (cm), H is the total pressure head including the gravitational and pressure head (Warrick, 2002).

Continuity equation For one-dimensional vertical water flow the continuity equation is written as: 휕휃 휕푞 (2-6) = − − 푆 휕푡 휕푧 (ℎ) Where the left side of the expression represents the changes in volumetric water content (cm³/cm³) per time (day) and the right side of the expression refers to water flux changes (cm) over distances (cm) per time (day) (Warrick, 2002; Hendriks, 2010; Radcliffe and Simunek, 2010). The

S(h) is the sink term which is a function of the negative pressure head (cm³/cm³/day) (Radcliffe and Simunek, 2010).

Richards equation For one-dimensional water flow the Richards model is written as (Warrick, 2002): 휕휃 휕 휕ℎ (2-7) = [퐾 ( + 1)] + 푆 휕푡 휕푧 (ℎ) 휕푧

Water movement on crops On the soil the roots absorb water when contacting soil moisture (Taiz et al., 2015). The contact area of the roots with the soil is increased by the roots’ hairs (Taiz et al., 2015). The highest root 16

State of the art absorption takes place near the root tips (Taiz et al., 2015). The root water uptake decreases the soil moisture surrounding the roots and this makes the water flow from longer distances (Taiz et al., 2015). This water flux is higher when the soil moisture content is high, however the soil water flux is reduced with the increasing root water uptake owing to the decrease in soil hydraulic conductivity (Blume et al., 2016). From the roots, water moves through the xylem to the mesophyll cell walls of the leaves from where the water evaporates into the leaves’ air spaces (Taiz et al., 2015). This water vapor is transferred to the atmosphere through the stomata (Taiz et al., 2015). The stomata are small apertures in the leaves which allow the external (atmosphere) and internal (leaves) communication of the crops (Taiz et al., 2015). These orifices are surrounded by two guard cells which regulate the aperture of the pores according to the water status of the crops, water status of soil and weather conditions, such as temperature, pressure deficit, relative air humidity and wind (Chang, 2009). Stomata are concentrated on the lower part of big leaves from trees (Chang, 2009). Although in annual crops such as corn and grasses (Chang, 2009) the stomata are proportionally located in the upper and lower parts of the leaves. Circa 20,000 stomata are located in every cm² of leaves (Chang, 2009). The stomata aperture is essential for photosynthesis. Nevertheless, the loss of water is much higher than the uptake of CO2 (Taiz et al., 2015). In C3 crops, 400 water molecules are lost to each

CO2 molecule gained, whereas for C4 crops this number is 150 (Taiz et al., 2015). The CO2 and water vapor exchange occurs by diffusion. Diffusion is the natural movement of substances due to differences in concentration (Taiz et al., 2015). Gas exchange is essential for crop production, because photosynthesis is responsible for the synthesis of carbohydrates (Chang, 2009). These carbohydrates are produced in the leaves and distributed within the crop structure through the phloem (Taiz et al., 2015). Therefore, transpiration is associated with dry mass production (Chang, 2009).

Water status on the soil Water status on the soil helps to explain physical, chemical and biological processes (Topp and Ferré, 2002). Water status in the soil is evaluated by water amount and soil water potential (Lal and Shukla, 2004). Moisture volume can be determined using direct or indirect methods (Lal and Shukla, 2004). The most common direct method is using thermogravimetry (Lal and Shukla, 2004). This method determines the water mass by evaporating the soil moisture in an oven under 105 oC up to constant weight (Lal and Shukla, 2004). The evaporated water can then be associated with the dry weight of the samples and provide information about the water content by mass or by volume. Water content by volume is estimated when the sample volume or bulk density is known (Lal and Shukla, 2004). The moisture volume is widely used due to the ease of associating it with height (mm, cm).

17

State of the art

Indirect methods determine the induced changes in the water-soil properties (Lal and Shukla, 2004). These methods use time domain reflectometry, neutron thermalization and capacitance devices (Radcliffe and Simunek, 2010). Time domain reflectometry (TDR) relies on electrical magnetic pulses applied to the soil using two or three metal conductors. The travel time of the electromagnetic wave applied with the instrument is associated with the relative permittivity, εr, of the substances. The relative permittivity (dimensionless) of water is 80, for air it is 1 and for soil minerals it ranges from 5 to 7 (Lal and Shukla, 2004). Permittivity, ε, is a measure of soil resistance to an electrical impulse. The relative permittivity associates the permittivity of a determined medium to a vacuum or free space, which presents the lowest values, 1 (Lal and Shukla, 2004). Therefore, the velocity of the electromagnetic wave is inversely associated with relative permittivity (Radcliffe and Simunek, 2010). Neutron thermalization uses a neutron meter placed within soil access tubes (Radcliffe and Simunek, 2010). This device delivers radioactive elements, such as Americium and Berylium, with high speed neutrons which cross the access tube. The speed of the neutrons is reduced according to the moisture content of the medium (Radcliffe and Simunek, 2010). Capacitance devices also use an electromagnetic pulse and permittivity, however the frequency domain reflectometry (FDR) devices estimate the resonance frequency of the soil (Radcliffe and Simunek, 2010). As the permittivity is affected by salt content and organic matter, the devices should be calibrated according to specific uses (Lal and Shukla, 2004). The indirect methods to determine water content in soils are non-destructive, provide in situ measurements and can be used continuously for measurements. Gravimetric measurements, in contrast are time consuming and are more used as a reference method to calibrate indirect methods (Topp, 2003). In frozen soil the TDR and FDR may represent the unfrozen moisture because ice has a similar level of permittivity as dry soil (Watanabe and Wake, 2009; Smith and Tice, 1988). The energy status of the moisture can be evaluated through kinetic (velocity) or position energy (Radcliffe and Simunek, 2010). As the kinetic energy of the soil moisture is neglected, the energy status of the soil moisture is evaluated due to its position in the field (Radcliffe and Simunek, 2010). Water potential includes different forces that act on the water energy status, such as matric (adsorption and cohesion), gravity, pressure and salinity forces (Radcliffe and Simunek, 2010). Gravitational potential refers to the gravity portion of the total water potential (Radcliffe and Simunek, 2010). The matric potential indicates the adhesive and cohesive forces which decrease the water potential in relation to the atmospheric free water (Radcliffe and Simunek, 2010). Osmotic pressure refers to the salt concentration and pressure potential refers to the above free water pressure on the soil moisture (Radcliffe and Simunek, 2010). Adhesion considers the water and solid bonds (Brady and Weil, 2014). Whereas the cohesive forces are related to the hydrogen bonds between water molecules (Radcliffe and Simunek, 2010). Adhesion and cohesion are a result of the bipolar

18

State of the art water properties (Brady and Weil, 2014). The polarity of water molecules is due to the hydrogen electropositive side and oxygen electronegative side (Brady and Weil, 2014). The association between the water content and the energy status (matric potential) of the water, represents the water retention curve (Radcliffe and Simunek, 2010), also known as the pF curve (Blume et al., 2016) or soil moisture characteristics (Lal and Shukla, 2004). The pF values from the water retention curve correspond to the logarithm of the matric potential in the water column, hPa, i.e., 1 hPa = 1 cm water column (Blume et al., 2016). This curve is one of the main parameters in water flow processes (Radcliffe and Simunek, 2010). The water retention curve is unique to each soil or substrate because it represents the water flow within the specific minerals, pore arrangements and pore connectivity (Lal and Shukla, 2004). The water retention curve can be measured in the laboratory using pressure and suction plates (Lal and Shukla, 2004). The measurements in the laboratory are then used to adjust the van Genuchten (1980) water retention model using RETention Curve (RETC) program. The van Genuchten water retention curve model was published in the 1980´s and has since been widely applied in water flow studies (Patil and Singh, 2016).

휃 − 휃푟 1 (2-8) 푆푒 = = 푛 1 − 1 / 푛 휃푠 − 휃푟 [ 1 + ( 훼 . ℎ ) ]

Where Se is the relative saturation, Ɵ is the volumetric water content (cm³/cm³), Ɵr is the residual water content (cm³/cm³), representing the water content where hydraulic conductivity approximates zero (Radcliffe and Simunek, 2010); Ɵs is the water content at soil saturation (cm³/cm³), also considered the total porosity if there is no entrapped air (Radcliffe and Simunek, 2010); h is the pressure head (cm); α is related to the air-entry at the saturated zone and is equal to the inverse dimension of matric potential, > 0 (cm-1) (Radcliffe and Simunek, 2010); n is associated with the pore size distribution (> 1) (Schaap and Leij, 2000) and is associated with the steepness (slope) of the water retention curve (van Genuchten, 1980). By combining the Mualen (1976) pore size model with the van Genuchten (1980) water retention curve, van Genuchten (1980) proposed an expression to estimate the unsaturated hydraulic conductivity. This expression is termed Mualem-van Genuchten model (Schaap and Leij, 2000).

푚 2 푙 1/푚 (2-9) 퐾(ℎ) = 퐾푠 .푆푒 [1 − (1 − 푆푒 ) ]

Where K(h) is the unsaturated hydraulic conductivity function (cm/day); Se is the effective saturation (dimensionless); Ks is the saturated hydraulic conductivity (cm/day); and l is the pore- connectivity parameter, which is 0.5 for most of the soils (Mualem, 1976); m is a fitting parameter which is associated with n, both dimensionless, in which m = 1 - 1 (Simunek et al., 2013). The n unsaturated hydraulic conductivity is an imperative parameter in vadose hydrological studies (Schaap and Leij, 2000).

19

State of the art

Measuring the water retention curve, saturated and unsaturated hydraulic conductivity methods are costly and time consuming (Schaap et al., 2001). In addition, reliable data are difficult to measure due to the high variability of soil properties and crop growth in the field (van Genuchten, 1992). Therefore, pedotransfer functions can be used to estimate soil hydraulic properties and model fitting parameters (Patil and Singh, 2016). Pedrotransfer functions are indirect methods for estimating soil properties from less labor-intensive measurements, such as bulk density and texture (Patil and Singh, 2016). Schaap et al. (2001) developed Rosetta pedotransfer function. The Rosetta pedotransfer function is based on 2134 soil samples from temperate and subtropical climates (Schaap et al., 2001). The Rosetta pedotransfer model considers 5 different hierarchical input data to predict the water retention parameters and saturated hydraulic conductivity (Schaap et al., 2001). The first input data model considers the soil texture class, the second includes the rate of sand, silt and clay from the fine earth, the third considers the bulk density, the fourth includes the water content at 33 kPa and the fifth includes in addition the water content at 1500 kPa (Schaap et al., 2001). Further pedotransfer functions for tropical regions can be found in Hodnett and Tomasella, (2002), Patil and Singh (2016), and Saxton and Rawls (2006). The water retention curve informs on the pore size distribution, available and non-available moisture and drainage water (Horton et al., 2016). Total available water refers to the water stored from field capacity to permanent wilting point (Allen et al., 1998). The unavailable water is the water below the limits of root absorption (permanent wilting point). Field capacity represents the maximum moisture the substrate can hold against gravity. The field capacity can be measured in the field by ponding water in a few square meters and measuring the water content up to constant value. Radcliffe and Simunek (2010) highlight that the hydraulic conductivity approximates to zero at field capacity. At the permanent wilting point water is strongly bonded to the particles’ surface and is not available for root uptake and hence the crops wilt (Horton et al., 2016). The water retention curve is affected by bulk density, total pore volume, grain size distribution and structure. Bulk density shows the ratio between wet or dry soil mass and determined volume (Horton et al., 2016). Bulk density is measured using core method, clod method and excavation method (Shukla, 2014). The core method uses undisturbed samples collected with metal rings with a known volume (Shukla, 2014). The clod method uses intact samples from the soil (aggregate). The clod is firstly immersed in wax and then water (Shukla, 2014). The change in the water volume is recorded and registered as the clod volume (Shukla, 2014). Then the clod is oven dried up to a constant weight (Shukla, 2014). For the excavation method, the soil is removed from the field and oven dried. Then the excavated space is isolated with a plastic canvas and filled with water. The dried mass and the water volume is used to estimate the bulk density (Shukla, 2014). Total pore volume represents the air space of the soils or substrates. Total pore volume can be estimated using the bulk density and particle density or by saturation when there is no entrapped air in the samples. 20

State of the art

However, the water saturation of the samples provides between 5 and 10 % lower values than the current total pore volume due to entrapped or dissolved air (van Genuchten et al., 1991). Grain size distribution is one of the main parameters in water flow studies because it contributes to the specific surface area, where water and nutrients interact. From the grain size distribution, it is possible to determine the soil texture, sand, silt or clay. Large surface areas and pores with smaller diameters provide a high-water retention capacity in clay soil (Taiz et al., 2015). Whereas in sandy soils with particle´s larger than 1 mm, a higher water drainage rate is found (Taiz, et al., 2015). Structure represents the arrangement of the minerals on the flow domains (Shukla, 2014). Soil structure can be classified as granular or crumby (spherical), blocky, columnar or prismatic, platy and massive (Shukla, 2014). Another phenomenon that affects water retention curves is hysteresis. Hysteresis occurs when the water retention curve differs by drying or wetting measures (Shukla, 2014). In the field this phenomenon is found in precipitation evaporation processes (Shukla, 2014).

2.1.5.4 Factors that determine the evapotranspiration rate The main parameters that affect evapotranspiration are weather, crops and soil parameters. Weather parameters that affect evapotranspiration are, solar radiation, air temperature, relative air humidity and wind speed.

2.1.5.4.1 Solar radiation The mean short wave radiation which arrives on the top of the earth´s atmosphere (extraterrestrial solar radiation) is 1366 W/m² (1 Watt is equal to 1 Joule per second) (Shuttleworth, 2012). From this circa 28 % is reflected from the clouds, 16 % is absorbed in water vapor, oxygen, ozone, and CO2 molecules in the atmosphere, 11 % is scattered by particles in the air out of the atmosphere and 26 % is scattered on the earth´s surface (Radcliffe and Simunek, 2010). Therefore, 19 % of the sun’s radiation reaches the earth´s surface directly. Also, accounting for 26 % of radiation scattered on the earth’s surface this rate increases to 45 % (Radcliffe and Simunek, 2010). This form of solar radiation reaching the earth is known as global solar radiation (Radcliffe and Simunek, 2010). From global solar radiation, one can determine the net downward radiation (Radcliffe and Simunek, 2010). The net downward radiation accounts for albedo, the long waves emitted outwards from the earth´s or canopy´s surface and long waves emitted downwards from the clouds (Radcliffe and Simunek, 2010). Albedo refers to the reflection rate of the short waves from the earth´s surface (Lal and Shukla, 2004). The reflection rate of solar radiation varies according to the surface, latitude and angle of the sun (Lal and Shukla, 2004). The albedo changes according to the soil color, i.e., light colored soil reflects more than dark colored soil (Lal and Shukla, 2004). Moreover, dry soils reflect more than wet soils (Lal and Shukla, 2004). The albedo increases for high latitudes which 21

State of the art are more distant from the equator (Lal and Shukla, 2004). The mean albedo on earth is 0.36 (Radcliffe and Simunek, 2010), forests reflect from 5 to 20 % of solar radiation (Lal and Shukla, 2004), fresh snow 70 % (Radcliffe and Simunek, 2010), open water surfaces 8 %, dirty old snow 40 %, and bare soil and agricultural crops 23 % (Shuttleworth, 2012). Anything that has a temperature above zero degrees emits energy (Shuttleworth, 2012). The sun has a temperature of circa 6000 K (5726.85 oC) and emits short waves or solar radiation, whereas the earth has a temperature of 290 K (16.85 oC) and emits longwave radiation. Wavelength refers to the distance between waves peaks within the repeating wave cycle (Henderson, 2017). Short waves are characterized by the high energy and wavelengths of < 4 μm (Wallace and Hobbs, 2006). Whereas long waves have a length of > 4 μm (Wallace and Hobbs, 2006). From the soil surface net radiation, the latent heat, sensible heat and soil heat transfer is estimated (Shuttleworth, 2012; Hendriks, 2010). The latent heat is the heat removed from the earth´s surface as part of the evaporation process or toward the soil surface if there is condensation (Shuttleworth, 2012; Hendriks, 2010). Sensible heat is the heat energy that is transferred from the earth´s surface to the atmosphere when the air above the surface is warmer than the overlying air (Shuttleworth, 2012; Hendriks, 2010). The soil heat flux is the heat energy that is transmited by conduction into the soil profile (Shuttleworth, 2012). The soil heat flux is positive when the heat transfer is downwards, i.e. during the day, and negative when it is upwards, e.g., at night, that is why soil heat fluxes are generally neglected for periods ranging from 10 to 30 days in hydrological studies (Hendriks, 2010). The local radiation on the earth varies according to the latitude, time of the day, day of the year and cloudiness (Allen et al., 1998). The amount of radiation varies according to the time of the day as it is due to the rotation of the earth (day and night) and the day of the year due to the earth´s inclination (seasons) (Allen et al., 1998). The global solar radiation (direct and diffuse) can be measured using pyranometers, radiometers and solarimeters (Allen et al., 1998). The solar radiation sensors are placed within glass protectors to avoid interference with rain and heat exchange (Shuttleworth, 2012).

2.1.5.4.2 Air temperature Air temperature is a measure of air molecule motion (Burt, 2012). This motion increases with available energy from the suns’ radiation and from the earth´s surface energy emission (Shuttleworth, 2012). On the earth´s surface, sensible heat is transformed into latent heat if there is moisture available to evapotranspirate (Allen et al, 1998). The evaporation of water at 20 oC consumes 2.45 MJ/kg and this is termed the latent heat of vaporization (Allen et al., 1998). Air temperature can be measured using thermometers (Burt, 2012). Thermometers contain a thin glass tube with a liquid which expands with temperature, i.e., mercury, the increase or decrease in air 22

State of the art temperature changes the length of the liquid within the glass tube (Burt, 2012). Presently, more advanced temperature sensors are based on electrical resistance and the mechanical expansion of specific materials, such as metals (Burt, 2012). For evapotranspiration studies, air temperature sensors should be placed at 2 m height (Allen et al., 1998).

2.1.5.4.3 Relative air humidity Relative air humidity is defined as the ratio of the actual vapor pressure in relation to the saturated vapor pressure (Monteith and Unsworth, 1990). The actual vapor pressure consists of the current moisture available in the air. Whereas the saturated vapor pressure considers the maximum air humidity if the air would be saturated at the same temperature (Allen et al., 1998). Another moisture vapor measure to be considered is dew point. Dew point represents the temperature at which the air vapor pressure becomes saturated (Monteith and Unsworth, 1990). The relative air humidity follows the inverse trend of solar radiation and air temperature, it is lower during the day and higher at night. This is because during the day there is more available energy and a higher capacity of the air to retain moisture. The relative air humidity changes according to the wind, precipitation, solar radiation and temperature.

2.1.5.4.4 Wind speed Wind speed is a mean value estimated over a time interval (Allen et al., 1998). The direction of the wind is also normally recorded (Allen et al., 1998). Anemometers are commonly used to measure wind speed (Allen et al., 1998). As surface friction reduces the wind speed, wind meters are placed at 10 m height in meteorology or up to 3 m height in agrometeorological studies (Allen et al., 1998), although for evapotranspiration studies wind meters are gerenally measured at 2 m height (Allen et al., 1998). If measurements are recorded at different heights, FAO recommends using a logarithm expression (Allen et al., 1998): 4.87 (2-10) 푢 = 푢 . 2 푧 푙푛(67.8 . 푧 − 5.42)

Where: u2 is the wind speed at 2 m height (m/s), uz is the wind speed (m/s) at a specific measured height above ground surface, z (m).

2.1.5.4.5 Soil The soil parameters that affect the potential crop growth and water fluxes also interfere in the evapotranspiration, such as salinity, pH, bulk density, soil moisture, soil texture, hydraulic conductivity, water retention curve, pore distribution, pore connectivity, soil color and soil temperature. Bulk density is associated with packing mineral particles at a determined volume (Horton et al., 2016). High soil densities mean low air space for root growth, water flow and water 23

State of the art retention. Hydrophobic soils reject water and have higher seepage rates. Hydrophilic soils have good contact with water and therefore can retain water. Hydraulic conductivity represents the capacity of a soil matrix to transmit water. Soil with a high hydraulic conductivity can transmit water at high speed, however, in this situation the seepage and the leaching of nutrients is high (Lal and Shukla, 2004). The soil texture is directly related to evapotranspiration because fine particles have a larger surface area and therefore a higher capacity to retain water. However, static analyses such as bulk density, particle density and soil texture are insufficient to conduct a study of the water retention capacity of soils. Therefore, dynamic studies such as saturated hydraulic conductivity, infiltration capacity and soil water retention curve must be studied. Soil moisture is particularly important because no evapotranspiration occurs if there is no moisture in the substrates to be transported to the atmosphere by the crops (Allen et al., 1998).

2.1.5.4.6 Crop Crop type, growth stage, leaf area index, root depth, root density, crop height, stomatal resistances (Shuttleworth, 2012), cuticle wax (Shuttleworth, 2012), plant density, plant nutrition, pest management interference with crop evapotranspiration (Allen et al., 1998).

2.1.5.4.7 Evapotranspiration concepts Due to the need to estimate evapotranspiration for an irrigation schedule, water management and hydrological studies, researchers defined three scientific terms for evapotranspiration studies, i.e., reference evapotranspiration, crop evapotranspiration and actual evapotranspiration (Goyal and

Harmsen, 2014; Allen et al., 1998). Reference evapotranspiration (ET0) refers to the evapotranspiration of a hypothetical grass crop at a height of 0.12 m, a fixed surface resistance of 70 sec/min, an albedo of 0.23 and no water limitations (Allen et al., 1998). Crop evapotranspiration

(ETc) is the evapotranspiration rate from a field crop under optimum conditions (Allen et al., 1998).

Whereas the actual evapotranspiration (ETa) refers to the quantity of water that is removed from the crops under non-standard environments or agronomic practices, such as fertilization and pest control (Allen et al., 1998).

2.1.5.5 Measurement and estimation of evapotranspiration Evapotranspiration can be measured or estimated (Abtew and Melesse, 2013). Measuring evapotranspiration is done mainly with lysimeters (Goyal and Harmsen, 2014). Additional evapotranspiration measurements can be performed using pan evaporation (Doorenbos and Pruitt, 1977), soil water depletion, water balance (Jensen et al., 1990), eddy correlation, bowen ratio and satellite-based methods (Abtew and Melesse, 2013). However, due to the high costs to install and run experiments to measure evapotranspiration (Bethune et al., 2008), several mathematical 24

State of the art expressions have been developed to estimate evapotranspiration (Abtew and Melesse, 2013; Tukimat et al., 2012). These mathematical expressions are classified into temperature, radiation or combination models (Abtew and Melesse, 2013; Liu et al., 2017). Temperature and radiation explain up to 80 % of reference evapotranspiration (Čadro et al., 2017). Temperature-based models are some of the oldest around (Xu and Singh, 2002) and are widely used because most weather stations measure temperature (Čadro et al., 2017). The radiation methods are generally based on energy balance (Xu and Singh, 2002) and on the conversion of energy on the water surface (Yeh, 2017). Temperature-based methods include the Blaney-Criddle method (Abtew and Melesse, 2013), the Blaney-Criddle modified method (Doorenbos and Pruitt, 1977), the Hargreaves and Samani method (Hargreaves and Samani, 1985) and the Thorntwaite method (Thorntwaite, 1948). Examples of the radiation methods are the Abtew method (Abtew, 1996), the Makkink method (Hendriks, 2010), the Priestley-Taylor method (Priestley and Taylor, 1972), the Turc method (Abtew and Melesse, 2013) and the FAO-24 Radiation method (Doorenbos and Pruitt, 1977). The combination methods are known as the energy balance method (Jensen et al., 1990), the Penman method (Penman, 1948), the mass transfer method (Harbeck, E., 1962), and the Penman-Monteith method (Allen et al., 1998). The evapotranspiration models were developed in specific regions (Lu et al., 2005), mainly in The United States and Western Europe (Xu and Singh, 2000; Jensen et al., 1990). Therefore, calibration is required to adjust the evapotranspiration models to the local weather data (Abtew and Melesse, 2013; Liu et al., 2017; Čadro et al., 2017). Local calibration can be performed using lysimeters or the standard FAO Penman-Monteith method (Liu et al., 2017; Allen et al., 1998). Calibration increases the reliability and accuracy of the estimations by using correction factors (Goyal and Harmsen, 2014).

2.1.5.5.1 Measurement of evapotranspiration

(a) Lysimeters Lysimeters are containers filled with disturbed or undisturbed soil and placed in the field (Goyal and Harmsen, 2014). Lysimter originates from the greek word “lysis” which means dissolved or movement and “metron”, meaning to measure (Goyal and Harmsen, 2014). Lysimeters can have different shapes (round, square or rectangular), can be made of different materials (concrete, plastic, fiberglass, steel), and have different sizes (Goyal and Harmsen, 2014). Lysimeters are widely used to study water balance components such as evapotranspiration and seepage (Aboukhaled et al.,

25

State of the art

1982). In addition, environmental aspects of the seepage properties, i.e., chemical composition, have also been investigated (Aboukhaled et al., 1982; Meissner et al., 2007). The first lysimeter was used in France in 1688 to study de evaporation of bare sand and grass (Aboukhaled et al., 1982) and the first lysimeter equipped with weighing devices was installed in Germany in 1906 by Conrad von Seelhorst (Aboukhaled et al., 1982). Presently, lysimeters are classified as weighing or non-weighing (Aboukhaled et al., 1982). Non-weighing lysimeters consider volumetric measurements of the water balance components, such as seepage and runoff (Goyal and Harmsen, 2014). The actual evapotranspiration from a field crop or grass in non- weighing lysimeters is estimated using the water balance expression (Aboukhaled et al., 1982):

퐸푇 = 푃 + 퐼 − 퐷 ± ∆푊 (2-11) Where ET is the evapotranspiration, P is precipitation, I is irrigation, D is the drainage and ΔW is the water storage change, in mm (Aboukhaled et al., 1982). The water storage change is positive when increasing and is negative when decreasing. Since seepage is a slow process, the evapotranspiration estimated using non-weighable lysimeter should be performed in long intervals (Goyal and Harmsen, 2014). The interval varies according to the size of the lysimeter, in which the initial and final moisture statuses are similar (Goyal and Harmsen, 2014). Weighing lysimeters are the most precise devices for measuring water balance components by mass (Goyal and Harmsen, 2014). Weighing lysimeters are equipped with load cells which determine the mass changes of the lysimeter vessel. Precise weighing lysimeters measure dew as well (Oberholzer et al., 2017; Hoffmann et al., 2016; Meissner et al., 2007). The interval of assessments is roughly an hour, making the intervals of assessments shorter than non-weighing lysimeters (Goyal and Harmsen, 2014). Precise measurements using lysimeters are achieved by surrounding the lysimeters with the same crop that is used within the lysimeter´s circumference (Goyal and Harmsen, 2014). Lysimeters must be deep enough to allow root growth and should have a representative area to account for the rim effect, as well as crop and soil variability (Goyal and Harmsen, 2014). However, issues arise in lysimeter measurements due to preferential flow in the lysimeters’ walls (Goyal and Harmsen, 2014). As the continuous measurements of lysimeters are costly and represent a specific area, precise measurements are used to calibrate evapotranspiration models (Liu et al., 2017).

(b) Pan evaporation Pan evaporation is widely used due to its simplicity and low cost (Lim et al., 2013). Pan evaporation consists of a circular container filled with water and exposed to environmental conditions (Shuttleworth, 2012). There are several types of evaporation containers (Bernardo et al., 2006). Variations of pan evaporation include changes in the installation height above the ground,

26

State of the art the use of a bird guard, square containers, and buried water reservoirs (Abtew and Melesse, 2013). However, the most known is the US Weather Bureau ´Class A´ pan (Shuttleworth, 2012; Bernardo et al., 2006). This container has a diameter of 121 cm and a depth of 25.5 cm (Doorenbos and Pruitt, 1977). The container is made of galvanized iron and placed on the field at 0.15 cm above the ground fixed on a wood structure (Shuttleworth, 2012). The pan’s water level must be 5 to 7.5 cm from the upper rim (Abtew and Melesse, 2013), to maintain a variation of 2.5 cm in the water level (Bernardo et al., 2006). The evaporation pans are often equipped with precipitation gauges and a weather station to observe meteorological conditions, such as air temperature and wind speed (Abtew and Melesse, 2013; Lim et al., 2013). The evaporation of the pan is evaluated using the following expression (Abtew and Melesse, 2013).

퐸푝푎푛 = 퐷푡−1 − 퐷푡 + 푅푓 − 퐿 ± 푒 (2-12)

Where Dt is current observed water depth and Dt-1 is the water level on the previous day; Rf is rainfall; L is water loss; and e is errors (Abtew and Melesse, 2013). Water loss in evaporation pans are due to animal consumption and errors including algae growth, accuracy of measurements and rain splash in and out of the evaporation pans (Abtew and Melesse, 2013; Linacre 1994). Bird guards decrease pan evaporation by circa 7 % (Abtew and Melesse, 2013). Evaporation measured in evaporation pans must be corrected with pan coefficients to estimate reference evapotranspiration (Abtew and Melesse, 2013), as follow:

퐸푇0 = 퐾푝 퐸푝푎푛 (2-13)

This correction is made using correction coefficients (Kp) which adjust the evaporation of the water pan to a vegetated area owing to the area of the pan, heating of the pan walls, air flow from the surrounding area (Shuttleworth, 2012), solar reflection of the surface, color of the pan and nocturnal evaporation due to water heating storage (Doorenbos and Pruitt, 1977). The reflection of the free water (water at atmospheric pressure and free of solutes) ranges from 5-8 % whereas from a vegetated surface it ranges from 20-25 % (Doorenbos and Pruitt, 1977). The general pan coefficient varies according to the relative air humidity, wind speed and position of the pan in relation to the surrounding field (grass or bare soil) and ranges from 0.40 to 0.85 (Doorenbos and Pruitt, 1977). By comparing the U.S. Class A pan evaporimeter with the estimation using the Penman equation (Penman, 1946), Linacre suggested a pan coefficient of 0.7 (Linacre, 1994).

(c) Soil water depletion The soil water depletion method estimates the evapotranspiration by evaluating the change in soil moisture between intervals. For this the soil moisture within the root zone is monitored using the gravimetric method or through moisture meter sensors (Jensen et al., 1990). The moisture content is evaluated 2 to 4 days after irrigation and repeated at 7 to 15 days or before the next 27

State of the art irrigation (Jensen et al., 1990). Then the evapotranspiration is estimated using the following expression (Jensen et al., 1990):

푊 ∑푛푟 (휃 − 휃 ) ∆푆 + 푅 − 푊 (2-14) 퐸 = 푒푡 = 푖=1 1 2 푖 푖 푒 푑 푡 ∆푡 ∆푡

Where Et is the evapotranspiration, nr is the number of soil layers in the root zone, ΔSi is the layer thickness in mm, θ1 is the first date volumetric water content and θ2 is the second date of volumetric water content in m³/m³, Re is rainfall in mm, Wd is the drainage studied soil layer in mm and Δt is the time interval between sampling dates (Jensen et al., 1990). This method has uncertainties due to the difficulties in measuring the drainage and upward moisture movement within the root zone (Jensen et al., 1990). The soil water depletion method considers generally a soil depth of 10 to 30 cm (Bernardo et al., 2006).

(d) Water balance This method accounts for inputs, outputs and moisture storage change in a system (Jensen et al., 1990). Water balance is studied to measure the evapotranspiration in lysimeters and in watersheds because both methods have a confined system (Jensen et al., 1990).

푃 + 퐼 ± 푅표 = 퐸푇 + 푃푅퐾 + 퐿 ± 훥푆푊 + 푒푟푟표푟 (2-15) Where P is the precipitation; I is irrigation; Ro is surface runoff; ET is actual evapotranspiration; PRK is deep percolation; L is lateral flow; ΔSW is change in soil water storage; error is the lack of balance in the measured terms (Hauser, 2009).

(e) Eddy covariance The eddy covariance studies the association between vertical wind speed and air humidity (Abtew and Melesse, 2013). When the wind speed and air humidity present a positive correlation, the moisture moves away from the evaluation site (Abtew and Melesse, 2013). And if the specific wind speed and humidity present negative values, there is a downward movement of dry air (Abtew and Melesse, 2013).

퐸 = 푤´푞´ = 1 ∑푁 (푤 − 푤) (푞 − 푞) (2-16) 푁 푖=1 푖 푖

Where E is the vapor flux, wi is the specific vertical wind speed, and qi is the specific air humidity at time i. 푤 and 푞 are mean specific wind speed and air humidity (Abtew and Melesse, 2013). The eddy covariance is evaluated using fast responding sensors (Abtew and Melesse, 2013). The device also requires high maintenance and does not guarantee continuous measurements (Abtew and Melesse, 2013).

28

State of the art

(f) Bowen ratio The Bowen method replaces the sensible heat in the energy balance equation by the Bowen ratio (Abtew and Melesse, 2013). The Bowen ratio is the ratio between sensible and latent heat flux (Abtew and Melesse, 2013). 푅 − 퐺 (2-17) 휆퐸 = 푛 1 + 훽

퐻 ∆푇 (2-18) 훽 = = 훾 휆퐸 ∆푒

Where λE is the latent heat flux, Rn is the net solar radiation, G is the soil heat flux, β is the Bowen ratio, H is the sensible heat, γ is the psychrometric constant, Δt is change in temperature, and Δe is the change in vapor pressure. The estimation of the Bowen ratio requires temperature and vapor pressure measurements from two heights above the surface (Abtew and Melesse, 2013).

(g) Satellite-based methods Satellite based methods can evaluate the surface temperature, albedo, radiation and heat fluxes between the soil surface and the atmosphere using maps (Abtew and Melesse, 2013). Satellite based models are advantageous as they evaluate the energy fluxes in a wide range of fields, such as rangelands, deserts, rivers and agricultural fields (Abtew and Melesse, 2013).

2.1.5.5.2 Estimation of evapotranspiration - temperature-based methods

(a) FAO-24 Blaney-Criddle method The original Blaney-Criddle method (Blaney and Criddle, 1950) was modified by the FAO (Doorenbos and Pruitt, 1977) to include additional weather parameters beyond temperature and daylight hours (Abtew and Melesse, 2013). The modified Blaney-Criddle method estimates the grass evapotranspiration by considering temperature, relative air humidity, daytime hours, site elevation and wind speed (Abtew and Melesse, 2013). This method considers the following expressions (Allen and Pruitt, 1986): 푒푙푒푣 (2-19) 퐸푇 = 푎 + 푏 (푝 (0.46 푇 + 8.13)) [1 + 0.1 ( )] 0 푎푣푔 1,000

Where ET0 is the estimated grass evapotranspiration (mm/day) for the studied time (month), Tavg is the mean temperature of the month (oC), p is the mean daily percentage of total annual daytime hours for the month and latitude, elev is the elevation above sea level in meters, the a and b are adjustment factors for the minimum relative air humidity (RHmin), sunshine hours (Nratio) and day-time wind speed (Uday) (Allen and Pruitt, 1986). 29

State of the art

푎 = 0.0043 . (푅퐻푚𝑖푛) − 푁푟푎푡𝑖표 − 1.41 (2-20)

푏 = 0.81917 − 0.0040922 (푅퐻푚𝑖푛) + 1.0705 (푁푟푎푡𝑖표) + (2-21) 0.065649 (푈푑푎푦) – 0.0059684 (푅퐻푚𝑖푛) (푁푟푎푡𝑖표) – – 0.0005967 (푅퐻푚𝑖푛)(푈푑푎푦) In areas where only 24-hrs of wind are available, daytime wind speed can be estimated as (Allen and Pruitt, 1986): 푈24 (푈푟푎푡𝑖표) (2-22) 푈푑푎푦 = 43.2 (1 + 푈푟푎푡𝑖표) Where U24 is the 24-hr wind speed in kilometers/day and Uratio is the ratio of daytime to night- time wind speeds (Allen and Pruitt, 1986). A value of 2.0 is suggested for Uratio where determinations of the day/night wind ratio are not available (Allen and Pruitt, 1986). The mean ratio of actual to possible sunshine hours can be estimated from solar radiation as follows (Allen and Pruitt, 1986): 푅 (2-23) 푁푟푎푡𝑖표 = 2.0 ( 푠 ) − 0.5 푅푎

Where Rs is the measured or estimated global solar radiation in mm/day, and Ra is the extraterrestrial short-wave solar radiation in mm/day (Allen and Pruitt, 1986).

(b) Hargreaves-Samani method The Hargreaves-Samani method (Hargreaves and Samani, 1985) can be implemented in studies where either the weather data are not accurate, or the only available data is the temperature (Hargreaves and Samani, 1985; Bos et al., 2009; Moeletsi et al., 2013). Although this method considers extraterrestrial radiation, it is classified as a temperature based method because the extraterrestrial data can be estimated (Abtew and Melesse, 2013).

0.5 퐸푇0 = 0.0023 × 0.408푅퐴 × (푇푎푣 + 17.8) × 푇퐷 (2-24)

Where ET0 is the reference evapotranspiration (mm/day); RA is the extraterrestrial radiation o (MJ/m²/day); Tavg is the average daily air temperature ( C), T = (Tmax + Tmin)/2; TD is the difference o between mean daily maximum and minimum temperature ( C), TD = (Tmax - Tmin). The 0.408 converts the radiation to evaporation in mm and the 0.0023 and 17.8 were obtained by fitting the equation to the measured reference evapotranspiration (Bos et al., 2009). The Hargreaves-Samani method was modified by Droogers and Allen (2002) to better fit the Penamn-Monteith equation (Bos et al., 2009). Droogers and Allen (2002) included the precipitation in the original Hargreaves-Samani method to decrease the differences with the Penamn-Monteith equation in very dry and very wet climates (Bos et al., 2009).

30

State of the art

0.76 퐸푇0 = 0.0013 × 0.408푅퐴 × (푇푎푣 + 17.0) × (푇퐷 − 0.0123푃) (2-25) Where P is precipitation (mm/month).

(c) Thornthwaite method The Thornthwaite model (Thornthwaite, 1948) estimates the monthly potential evapotranspiration using temperature values (Sentelhas et al., 2010). 10푇 푎 (2-26) 퐸푇 = 16 ( ) 푓표푟 0 0퐶 ≤ 푇 ≤ 26.5 0퐶 푝 퐼 For temperatures above 26 oC, Willmott et al. (1985), suggested the following expression:

2 0 퐸푇푝 = −415.85 + 32.24푇 − 0.43푇 푓표푟 푇 ≥ 26.5 퐶 (2-27)

Where ETp is the evapotranspiration for a month (mm/month); T is monthly mean air temperature (oC) and I and a are thermal indices, estimated using the following expressions (Sentelhas et al., 2010; Pereira and Pruitt, 2004):

퐼 = 12 (0.2 푇푎)1.514 푓표푟 푇푎 > 0 0퐶 (2-28)

푎 = 6.75 × 10−7 퐼3 − 7.71 × 10−5 퐼2 + 1.7912 × 10−2 퐼 + 0.49239 (2-29) Where Ta is the climatological normal of the annual temperature (oC). The monthly potential evapotranspiration is then converted to daily evapotranspiration using the following expression (Sentelhas et al., 2010): 퐸푇 푁 (2-30) 퐸푇 = 푝 0 30 12 Where N is the photoperiod (h) for a given day, circa 12 h (Sentelhas et al., 2010).

2.1.5.5.3 Estimation of evapotranspiration - radiation-based methods

(a) Abtew method The Abtew method (Abtew, 1996) was developed to estimate the lake evaporation, wetland evapotranspiration and potential evapotranspiration (Abtew and Melesse, 2013) using the following expression: 푅 (2-31) 퐸푇 = 퐾 푠 1 휆 Where ET is the daily wetland evapotranspiration or shallow open water evaporation or potential evapotranspiration (mm/day), Rs is the solar radiation (MJ/m²/day), λ is latent heat of water vaporization (MJ/kg), and k1 is a dimensionless coefficient (0.53).

31

State of the art

(b) Makkink method The Makkink method estimates the reference evapotranspiration using the following expression (Abtew and Melesse, 2013): 훥 푅 (2-32) 퐸푇 = 0.61 푠 (훥 + 훾)휆

Where ET is potential evapotranspiration (cm/day), Rs is the solar radiation in cal/cm²/day, Δ is the slope of the saturation vapor pressure (mb/oC), γ is the psychrometric constant (mb/oC), and λ is the latent heat of vaporization (cal/g). 4098 푒 (2-33) 훥 = 푠 (237.3 + 푇)2

o Where es (kPa) is the saturation vapor pressure and T is air temperature ( C) 푃 (2-34) 훾 = 0.0016286 휆 Where P (kPa) is the atmospheric pressure.

휆 = 2.501 − (0.00236 푇푠) (2-35) Where T (oC) is the surface temperature of water.

Lamb (2015) and Goyal and Harmsen (2014) describe the Makkink method in this sequence: 훥 (2-36) 퐸푇 = 푅 ( ) + 0.12 0 푠 훥 + 훾

Where ET0 is the potential evapotranspiration (mm/day), Rs is the radiation at the surface, expressed as equivalent evaporation (mm/day), Δ is the slope of the vapor pressure curve (kPa/oC), and γ is the psychrometric constant (kPa/oC).

Xu and Singh (2000) present and additional expression for the Makkink method, which is: Δ R (2-37) ET = 0.61 s - 0.12 0 Δ + γ 58.5

Where ET0 is the evapotranspiration (mm/day), Rs is the solar radiation (mm/day), Δ is the slope of the vapor pressure curve (mbar/oC), γ is the psychrometric constant (mb/oC). These values are estimated according to the equations:

훥 = 33.8639 [0.05904 (0.00738 푇 + 0.8072)7 − 0.0000342] (2-38)

퐶 푃 (2-39) 훾 (푚푏푎푟/ 표퐶) = 푝 0.622 휆

휆 (푐푎푙/푔) = 595 − 0.51 푇 (2-40)

32

State of the art

P = 1013 - 0.1055 EL (2-41) Where EL is the elevation above sea level (m), λ is latent heat (calories/gram), P is the o atmospheric pressure (mbar), Cp is the specific heat of the air (cal/g/ C), is circa 0.242. According to Hendriks (2010), the Makkink method is used by the Royal Dutch Meteorological Society (KNMI) and performs well in the Netherlands. Hendriks (2010) writes the Makkink method as: 1000 ∆ (2-42) 퐸 = 퐶 × × 푆 푀퐾 푀퐾 휌휆 ∆ + 훾 푡

Where EMK is the Makkink reference crop evapotranspiration (mm/day), CMK is the empirical factor (0.65 for humid climate), St is the incoming short wave radiation at the earth´s surface (MJ/m²/day), Δ is the gradient of the saturation vapor pressure curve (kPa/°C), ρ is the water density (1000 kg/m³), λ is the latent heat of vaporization (≈ 2.45 MJ/kg), γ is the psychrometric constant (≈ 0.067 kPa/°C), 1000/ρλ factor (dm³/MJ) to convert from MJ/m²/day to mm/day (Hendriks, 2010).

(c) FAO-24 Radiation method The FAO-24 Radiation method is originally written as (Doorenbos and Pruitt, 1977; Goyal and Harmsen, 2014):

푃퐸푇 = 푐 × (푊 × 푅푠) (2-43)

Where PET is the potential evapotranspiration for the considered period (mm/day), Rs is the solar radiation (mm/day), W is a correction factor which is associated with temperature and altitude, c is the correction factor which considers relative air humidity and wind speed. The W factor represents the fraction of the solar radiation used in evapotranspiration for different values of temperature and altitude (Melo and Fernandes, 2012). ∆ (2-44) 푊 = ∆ + 훾 Where Δ is the slope of the vapor pressure curve (kPa/oC), and γ is the psychrometric constant (kPa/oC).

The FAO radiation method can also be expressed as: ∆ (2-45) 퐸푇 = 푐 + 푐 ( 푅 ) 1 0 ∆ + 훾 푠

Where Rs is solar radiation (mm/day), c0 is a constant of -0.3 mm/day, c1 varies according to the mean relative air humidity and daytime wind speed (Melo and Fernandes, 2012; Xu and Singh, 2000).

33

State of the art

2 2 푐1 = 푎0 + 푎1 푅퐻 + 푎2 푈2 + 푎3 푅퐻 푈2 + 푎4 푅퐻 + 푎5 푈2 (2-46)

Where a0 = 1.0656; a1 = -0.0012795; a2 = 0.044953; a3 = -0.00020033; a4 = -0.000031508; a5 = -0.0011026.

(d) Priestley-Taylor method The Priestley-Taylor method (Priestley and Taylor, 1972) contains a model to estimate the reference evapotranspiration in areas with high water content (Xu and Singh, 2000). This model excluded the aerodynamic component from the Penman equation (1948) and the energy element was multiplied by a factor, α = 1.26 (Xu and Singh, 2000). The Priestley-Taylor method (Priestley and Taylor, 1972) is similar to the Makkink method (Hendriks, 2010) however, the solar radiation was replaced by net radiation (Abtew and Melesse, 2013), as follow: ∆ 푅 (2-47) 퐸푇 = 훼 푛 (∆ + 훾) 휆

ET is potential evapotranspiration (mm/day), Rn is the net radiation (cal/cm²/day), Δ is the slope of the saturation vapor pressure (mb/oC), γ is the psychrometric constant (mb/oC), λ is the latent heat of vaporization (cal/g), and α is an empirical coefficient, α = 1.26 (Xu and Singh, 2000; Abtew and Melesse, 2013). Hendriks (2010) writes the Priestley-Taylor expression as: 1000 ∆ (2-48) 퐸푇 = 퐶 × × (푅 − 퐺) 푃푇 푃푇 휌휆 ∆ + 훾 푛

Where ETPT is the reference crop evapotranspiration (mm/day), CPT is experimental factor (1.26 for a humid climate and 1.74 for an arid climate (dimensionless)), Δ is the gradient of the saturation 2 vapor pressure curve (kPa/°C), Rn is net radiation at the earth´s surface (MJ/m /day), G is the soil heat transfer (MJ/m2/day), ρ is the water density (1000 kg/m³), λ is the latent heat of vaporization (≈ 2.45 MJ/kg), γ is the psychrometric constant (≈ 0.067 kPa/°C), 1000/ρλ factor (dm³/MJ) to convert from MJ/m²/day to mm/day (Hendriks, 2010).

(e) Turc method The Turc method was developed under the climatic conditions of Western Europe (Xu and Singh, 2000). The Turc model considers: 푇 (2-49) 퐸푇 = 0.013 (푅 + 50) 푓표푟 푅퐻 ≥ 50 푇 + 15 푠

푇 50 − 푅퐻 (2-50) 퐸푇 = 0.013 (푅 + 50) (1 + ) 푓표푟 푅퐻 < 50 푇 + 15 푠 70

34

State of the art

o Where T is the average air temperature ( C), Rs is the total solar radiation (cal/cm²/day), and RH is the average relative air humidity (%) (Xu and Singh, 2000). Seiler and Gat (2007) report an additional Turc (1954) expression which considers long-term annual means of precipitation (mm/year) and air temperature (oC). 푃 (2-51) 퐸푇 = 푃 2 √0.9 + ( ) 300 + 25푇 + 0.05푇3 Where ET is the evapotranspiration (mm/year).

2.1.5.5.4 Estimation of evapotranspiration - combination methods

(a) Penman method The original Penman method was developed for open water surfaces and considered a radiation term associated with net energy flux and a second aerodynamic term which comprises the water vapor movement from the evaporating surface to the adjacent air (Bos et al., 2009). The Penman method and the FAO Penman method are no longer recommended (Bos et al., 2009), these methods were replaced by the standard FAO Penman-Monteith expression (Bos et al., 2009). The Penman method is written as follows: ∆ 푅 − 퐺 훾 (2-52) 퐸 = × 푛 + 퐸 0 ∆ + 훾 휆 Δ + 훾 푎

Where E0 is the open water evaporation rate (kg/m²s), Δ is the slope of the saturation vapor o pressure (kPa/ C), Rn is the net radiation (W/m²), G is the heat flux density into the water body o (W/m²), λ is the latent heat of vaporization (J/kg), γ is the psychrometric constant (kPa/ C), and Ea is the air evaporative power (kg/m² s) (Bos et al., 2009; Melo and Fernandes, 2012). The potential evapotranspiration in mm/day is obtained by multiplying resulting E0 (kg/m² s) by 86,400 seconds (Bos et al., 2009).

(b) FAO-24 Penman method The FAO modified the original Penman evaporation of free water to grass evaporation because grass, 0.08-0.15 m tall, is the common crop surrounding weather stations (Bos et al., 2009). Moreover, a grass cover has a short-wave reflection of circa 0.25 instead of 0.05 found on open water (Bos et al., 2009). The FAO-Penman method can be written as: ∆ 푅 훾 (2-53) 퐸푇 = 푐 [ × 86400 푛 + 2.7 푓(푢) (푒 − 푒 )] 푔 ∆ + 훾 휆 Δ + 훾 푧,푠푎푡 푧

Where ETg is the reference evapotranspiration rate (mm/day), c is a dimensionless adjustment factor, Rn is an energy flux density of net incoming radiation (W/m²), f(u) wind function; f(u) = 1 35

State of the art

+ 0.864u2, u2 is the wind speed at 2-m height (m/s), ez,sat – ed is the vapor pressure deficit (kPa), Δ is the slope of the saturation vapor pressure (kPa/oC), and γ is the psychrometric constant (kPa/oC) (Bos et al., 2009).

(c) The Penman method modified by Monteith The Penman-Monteith evapotranspiration expression can be used to estimate the evapotranspiration of any crop, if the aerodynamic and surface resistance parameters are known (Pereira and Alves, 2013; Allen et al., 1998).

(푒푠 − 푒푎) (2-54) Δ (푅푛 − 퐺) + 휌푎 푐푝 1 푟푎 퐸푇 = 푟 휆 Δ + 훾 (1 + 푠 ) 푟푎

Where ET is the evapotranspiration (mm/day), λ is the latent heat of vaporization (kg/m³), Rn -

G is the net balance of energy available at the surface (MJ/m²/day), (es - ea) is the vapor pressure deficit (kPa), ρa is the mean air density (kg/m³), cp is the specific heat of the air at a constant pressure (kJ/kg/oC), Δ is the slope of the saturation vapor pressure-temperature relationship at the mean o o temperature (kPa/ C), γ is the psychrometric constant (kPa/ C), rs represents the bulk surface resistance (sec/m), and ra is the aerodynamic resistance (sec/m) (Pereira and Alves, 2013). The aerodynamic resistance represents the resistances to moisture and heat exchange above the crop surface (canopy) (Allen et al., 1998). The aerodynamic resistance is estimated using the expression (Allen et al., 1998): 푧 − 푑 푧 − 푑 (2-55) ln [ 푚 ] 푙푛 [ ℎ ] 푧표푚 푧표ℎ 푟푎 = 2 푘 푢푧

Where ra is the aerodynamic resistance (sec/m), zm is the height of the wind measurements (m), zh is the height of humidity measurements (m), d is the zero-plane displacement height (m), zom the roughness length governing momentum transfer (m), zoh roughness length governing transfer of heat and vapor (m), k is von Karman´s constant, 0.41 (dimensionless), uz wind speed at height z (m/s). The surface resistances which are associated with soil and leaf resistances can be estimated using the expression:

푟1 (2-56) 푟푠 = 퐿퐴퐼푎푐푡푖푣푒

Where rs is the bulk surface resistance (sec/m), r1 is the bulk stomatal resistance of well- illuminated leaves (sec/m), LAIactive is the active leaf area index, i.e., the leaf area which contributes to heat and vapor transfer (m² leaf area/m² of soil surface). The LAI varies depending on the stage of crop growth, plant density and variety, and it ranges from 3-5 in mature crops (Allen et al., 1998). 36

State of the art

(d) FAO-56 Penman-Monteith method The FAO reference crop evapotranspiration based on the Penman-Monteith equation parametrized for grass with a crop height of 0.12 m, a surface resistance of 70 sec/m and an albedo of 0.23 is estimated using the following expression (Allen et al., 1998): 900 ( ) ( ) (2-57) 0.408 . ∆ . 푅푛 − 퐺 + 훾 . 푇 + 273 . 푢2 . 푒푠 − 푒푎 퐸푇0 = ∆ + 훾 . (1 + 0.34 . 푢2)

ET0 is the reference evapotranspiration (mm/day), Rn is net radiation at the crop surface (MJ/m2/day), G is soil heat flux density which is considered null for daily estimates (Sentelhas et 2 al., 2010) (MJ/m /day), T is mean daily air temperature at 2-m height (°C), u2 is wind speed at 2-m height (m/s), es is saturation vapor pressure (kPa), ea is actual vapor pressure (kPa), es - ea is saturation vapor pressure deficit (kPa), Δ is the slope of the vapor pressure curve (kPa/°C), and γ is the psychrometric constant (kPa/°C) (Bos et al., 2009). Pereira and Alves (2013) present the equation for estimating the reference evapotranspiration for grass and alfalfa, which is:

0.408 ∆ (푅푛 − 퐺) + 훾 퐶푛 푢2 (푒푠 − 푒푎) (2-58) 퐸푇0 = ∆ + 훾 (1 + 퐶푑 푢2)

Where u2 is the wind speed (m/sec). Cn and Cd are the numerator and denominator coefficients that differ with the crop reference and calculation time type (Pereira and Alves, 2013). For daily estimation of reference evapotranspiration of grasses, surface resistance is 70 sec/m, which leads to a Cn of 900 and a Cd of 0.34 (Pereira and Alves, 2013).

The alfalfa reference (ETr) is defined as the evapotranspiration from full cover, well-watered, actively growing alfalfa with a standard height of 50 cm (Pereira and Alves, 2013). In this case, the

Cn and Cd are 1600 and 0.38 respectively for daily calculations. The reference evapotranspiration from alfalfa may reach 1.1 to 1.4 times the ET0. In addition, the ETr may be considered the maximum evapotranspiration expected from well-watered vegetation (Pereira and Alves, 2013).

(d.1) Calculation steps of the FAO-56 Penman-Monteith method The weather parameters used to estimate the FAO reference evapotranspiration (Allen et al., 1998) are estimated with the following expressions shown below.

(d.1.1) Air temperature 푇 + 푇 (2-59) 푇 = 푚푎푥 푚푖푛 푚푒푎푛 2

37

State of the art

o o Where Tmean is the daily mean temperature ( C), Tmax is the daily maximum air temperature ( C), o Tmin is the daily minimum air temperature ( C) (Allen et al., 1998).

(d.1.2) Saturation vapor pressure of the air The saturation vapor pressure is estimated from the air temperature and the expression is written as (Allen et al., 1998): 17.27 푇 (2-60) 푒표(푇) = 0.6108 푒푥푝 [ ] 푇 + 237.3 Where e0(T) is the saturation vapor pressure at air temperature T (kPa), T is the air temperature (oC), exp[..] is 2.7183 (base of natural logarithm) raised to the power [..](Allen et al., 1998). On a daily basis, the saturation vapor pressure is estimated using maximum and minimum air temperature (Allen et al., 1998). 푒표(푇 ) + 푒표(푇 ) (2-61) 푒 = 푚푎푥 푚푖푛 푠 2

Where es is the saturation vapor pressure (kPa) (Allen et al., 1998).

(d.1.3) Actual vapor pressure of the air The daily actual vapor pressure is similar to the saturated vapor pressure at the dew point temperature (Allen et al., 1998). The dew point temperature is the temperature in which the air becomes saturated (Allen et al., 1998).

표 17.27 푇푑푒푤 (2-62) 푒푎 = 푒 (푇푑푒푤) = 0.6108 exp [ ] 푇푑푒푤 + 237.3 The actual vapor pressure can be estimated using daily values of relative air humidity (Allen et al., 1998).

표 푅퐻푚푎푥 표 푅퐻푚푖푛 (2-63) 푒 (푇푚푖푛) + 푒 (푇푚푎푥) 푒 = 100 100 푎 2

o Where ea is the actual vapor pressure (kPa), e (Tmin) is the saturation vapor pressure at a daily o minimum air temperature (kPa), e (Tmax) is the saturation vapor pressure at daily maximum air temperature (kPa), RHmax is the maximum relative air humidity (%), RHmin is the minimum relative air humidity (%) (Allen et al., 1998). The relative air humidity measured early in the morning and early afternoon corresponds to the minimum and maximum relative air humidity respectively (Bos et al., 2009). For daily maximum relative air humidity and minimum air temperature the actual vapor pressure is written as:

38

State of the art

푅퐻 (2-64) 푒 = 푒표 (푇 ) 푚푎푥 푎 푚푖푛 100 For daily minimum relative air humidity and maximum air temperature the actual vapor pressure is estimated using the expression: 푅퐻 (2-65) 푒 = 푒표 (푇 ) 푚푖푛 푎 푚푎푥 100 For mean daily relative air humidity the actual vapor pressure is obtained by the equation: 푅퐻 푒표(푇 ) + 푒표 (푇 ) (2-66) 푒 = 푚푒푎푛 [ 푚푎푥 푚푖푛 ] 푎 100 2

(d.1.4) Psychrometric constant The psychrometric constant is the ratio of specific heat of moisture air at a constant pressure to latent heat of vaporization (Allen et al., 1998). The psychrometric constant is shown as: 푐 푃 (2-67) 훾 = 푝 = 0.665 푥 10−3 푃 휖 휆 Where γ is the psychrometric constant (kPa/oC), P is the atmospheric pressure (kPa), λ is the -3 latent heat of vaporization, 2.45 (MJ/kg), cp is the specific heat at constant pressure, 1.013×10 (MJ/kg/oC), ε is the ratio molecular weight of water vapor/dry air = 0.622 (Allen et al., 1998). The specific heat at constant pressure, cp, represents the amount of energy required to increase the temperature of a unity mass of air by one degree at constant pressure (Allen et al., 1998).

(d.1.5) Latent heat of vaporization The latent heat of vaporization represents the energy needed to evaporate the water at constant pressure (Allen et al., 1998). This value changes slightly according to the temperature, however for an air temperature of 20 oC, it has a value of 2.45 MJ/kg (Allen et al., 1998).

휆 = 2.501 − (2.361 × 10−3)푇 (2-68) Where λ is the latent heat of vaporization (MJ/kg), and T is the air temperature (oC)

(d.1.6) Atmospheric pressure The atmospheric pressure refers to the pressure exerted by the air extending from the earth´s surface to the outer limits of the atmosphere (Burt, 2012). The atmosphere is densest at the earth’s surface and decreases with elevation (Burt, 2012). In evapotranspiration studies the atmospheric pressure is given by:

39

State of the art

293 − 0.0065 푧 5.26 (2-69) 푃 = 101.3 ( ) 293 Where P is the atmospheric pressure (kPa), z is the elevation above sea level (m) (Allen et al., 1998).

(d.1.7) Slope of the saturation vapor pressure-temperature curve The slope of the relationship between saturated vapor pressure and temperature is given by (Allen et al., 1998): 17.27 푇 4098 [0.6108 푒푥푝 ( )] (2-70) Δ = 푇 + 237.3 (푇 + 237.3)2 Where Δ is the slope of saturation vapor pressure at a mean air temperature (kPa/oC), T is the mean air temperature (oC), exp [..] is the 2.7183 (base of natural logarithm raised to the power of [..] (Allen et al., 1998).

(d.1.8) Wind speed at 2-m height The wind speed changes with the height aboveground (Bos et al., 2009). For measurements performed above 2-m height and above short grass cover, the wind speed is adjusted using the expression (Bos et al., 2009): 4.87 (2-71) 푢 = 푢 . 2 푧 푙푛(67.8 . 푧 − 5.42)

Where: u2 is the wind speed at 2 m height (m/s), uz is the wind speed (m/s) at a specific measured height above the ground surface, z (m) (Allen et al., 1998).

(d.1.9) Net radiation For evapotranspiration studies the net radiation represents the energy available to evaporate the moisture and heat the surface air (Bos et al., 2009). The net radiation is given by:

푅푛 = 푅푛푠 − 푅푛푙 (2-72)

Where Rn is the net radiation ((MJ/m²/day), Rns is the net short wave radiation (MJ/m²/day), Rnl is the net outgoing long-wave radiation (MJ/m²/day) (Bos et al., 2009). The net short wave radiation considers the energy that arrives on the surface and the reflected energy (albedo), as follows (Allen et al., 1998):

푅푛푠 = (1 − 훼)푅푠 (2-73)

40

State of the art

Where Rns is the net solar or shortwave radiation (MJ/m²/day), α is the albedo or canopy reflection coefficient, Rs is the incoming solar radiation (MJ/m²/day) (Allen et al., 1998). For the FAO reference evapotranspiration, the albedo is 0.23 (Allen et al., 1998). The net outgoing long-wave radiation can be estimated using the following expression:

4 4 푇푘 푚푎푥 + 푇푘 푚푖푛 (2-74) 푅 = 훼 푓 (0.34 − 0.14 √푒 ) 푛푙 푐푑 푎 2

Where Rnl is the net outgoing long-wave radiation (MJ/m²/day), α is the Stefan-Boltzmann -9 4 constant (4.901 × 10 MJ/K /m²/day), fcd is a cloudiness constant (uniteless), and limited to 0.05 ≤ fcd ≤ 1, ea is the actual vapor pressure (ea) (kPa), TK max is the maximum absolute daily temperature o (K) (K = C + 273.15), TK min is the minimum absolute daily temperature (K) (Bos et al., 2009).

(e) Mass transfer method The mass transfer model focuses on mass transfer and aerodynamic terms and does not consider the energy required to evaporate the moisture from the free water surface (Abtew and Melesse, 2013). The mass transfer model can be written as (Abtew and Melesse, 2013):

퐸 = (푒푠푠 − 푒푑푑) 푓(푢) (2-75)

Where E is the evaporation per unit of time, ess is vapor pressure on the evaporating surface, edd is the vapor pressure in the atmosphere above, and f(u) is a function of the wind speed (Abtew and Melesse, 2013). Haude´s model (Haude, 1955) uses a similar expression and is very popular in Germany (Loos et al., 2007). Seiler and Gat (2007) higlight that the Haude´s model has been applied in arid regions as well (Haude, 1959). The Haude´s model is written as:

퐸푇푝퐻푎푢푑푒 = χ . (푒푠 − 푒푎)2 푝.푚. (2-76)

Where χ is the Haude factor according to the specific vegetation surface and es-e is the vapor pressure deficit in hPa for the temperature at 2 p.m. (Loos et al., 2007; Häckel, 1999) measured at 2-m height (Seiler and Gat, 2007). Table 2-1 presents an overview of the evapotranspiration measurements and models.

41

State of the art

Table 2-1: Measurements and mathematical expressions to estimate reference evapotranspiration

Method Equation Reference

ET measurements

Non-weighable lysimeter 퐸푇 = 푃 + 퐼 − 퐷 ± ∆푊 Aboukhaled et al. (1982) Class A Pan Doorenbos and Pruitt 퐸푇0 = 퐾푝 퐸푝푎푛 (1977) Soil water depletion 푊 ∑푛푟 (휃 − 휃 ) ∆푆 + 푅 − 푊 Jensen et al. (1990) 퐸 = 푒푡 = 푖=1 1 2 푖 푖 푒 푑 푡 ∆푡 ∆푡 Water balance 푃 + 퐼 ± 푅표 = 퐸푇 + 푃푅퐾 + 퐿 ± 훥푆푊 + 푒푟푟표푟 Hauser (2009)

Eddy covariance 푁 Abtew and Melesse 1 퐸 = 푤´푞´ = ∑(푤 − 푤) (푞 − 푞) (2013) 푁 푖 푖 푖=1 Bowen ratio 푅 − 퐺 Abtew and Melesse 휆퐸 = 푛 1 + 훽 (2013)

Temperature-based methods

FAO-24 Blaney-Criddle 푒푙푒푣 Doorenbos and Pruitt method 퐸푇0 = 푎 + 푏 (푝 (0.46 푇푎푣푔 + 8.13)) [1 + 0.1 ( )] (1977) 1,000

Hargreaves-Samani 퐸푇 = 0.0023 × 0.408푅 × (푇 + 17.8) × 푇퐷0.5 Hargreaves and Samani method 0 퐴 푎푣 (1985) Modified Hargreaves- 0.76 Droogers and Allen 퐸푇0 = 0.0013 × 0.408푅퐴 × (푇푎푣 + 17.0) × (푇퐷 − 0.0123푃) Samani method (2002) Thornthwaite method 10푇 푎 Thornthwaite (1948) 퐸푇 = 16 ( ) 푓표푟 0 0퐶 ≤ 푇 ≤ 26.5 0퐶 Willmott et al. (1985) 푝 퐼

2 0 퐸푇푝 = −415.85 + 32.24푇 − 0.43푇 푓표푟 푇 ≥ 26.5 퐶

Radiation-based methods

Abtew method 푅 Abtew (1996) 퐸푇 = 퐾 푠 1 휆 Makkink method 훥 Goyal and Harmsen 퐸푇 = 푅 ( ) + 0.12 0 푠 훥 + 훾 (2014) FAO-24 Radiation Doorenbos and Pruitt 푃퐸푇 = 푐 × (푊 × 푅푠) method (1977) Priestley-Taylor method ∆ 푅 Xu and Singh (2000) 퐸푇 = 훼 푛 (∆ + 훾) 휆 Abtew and Melesse (2013) Turc method 푇 Xu and Singh (2000). 퐸푇 = 0.013 (푅 + 50) 푓표푟 푅퐻 ≥ 50 푇 + 15 푠 푇 50 − 푅퐻 퐸푇 = 0.013 (푅 + 50) (1 + ) 푓표푟 푅퐻 < 50 푇 + 15 푠 70

Combination methods Penman method ∆ 푅 − 퐺 훾 Bos et al. (2009) 퐸 = × 푛 + 퐸 0 ∆ + 훾 휆 Δ + 훾 푎 42

State of the art

FAO-24 Penman method ∆ 푅 훾 Bos et al. (2009) 퐸푇 = 푐 [ × 86400 푛 + 2.7 푓(푢) (푒 − 푒 )] 푔 ∆ + 훾 휆 Δ + 훾 푧,푠푎푡 푧 Penman-Monteith Pereira and Alves (2013) (푒푠 − 푒푎) method Δ (푅푛 − 퐺) + 휌푎 푐푝 1 푟푎 퐸푇 = 푟 휆 Δ + 훾 (1 + 푠 ) 푟푎 FAO-56 Penman- 900 Allen et al. (1998) 0.408 . ∆ . (푅 − 퐺) + 훾 . . 푢 . (푒 − 푒 ) Monteith method 푛 푇 + 273 2 푠 푎 퐸푇0 = ∆ + 훾 . (1 + 0.34 . 푢2) Mass transfer method Abtew and Melesse 퐸 = (푒푠푠 − 푒푑푑) 푓(푢) (2013)

FAO Penman-Monteith method From all the mathematical expressions available to estimate reference evapotranspiration (Table 2-1) the Penman-Monteith equation is recommended by the Food Agricultural Organization (FAO) of the United Nations since 1990 as a standard method due to its reliability in different climate conditions (Allen et al., 1998). The Penman-Monteith equation considers temperature, solar radiation, relative air humidity, wind speed, aerodynamics and surface resistances (Lamb, 2015). Surface resistances are related with water vapor resistances through stomata, leaf area and soil (Goyal and Harmsen, 2014; Allen et al., 1998). Whereas the aerodynamic resistances are associated with wind frictions above the crops (Goyal and Harmsen, 2014; Allen et al., 1998). These resistances have dimensions of time per unit length (Goyal and Harmsen, 2014).

Crop evapotranspiration (ETc) is obtained by the ET0-Kc two-steps approach (Allen et al., 1998; Goyal and Harmsen, 2014). Crop coefficients adjust the reference evapotranspiration for specific crops and therefore integrate differences in crop height, crop growth and crop populations (Goyal

and Harmsen, 2014). The actual evapotranspiration (ETa) can either be estimated correcting the crop coefficients for all the environmental stresses or by adjusting crop evapotranspiration with

crop-water stress coefficients (Ks). Crop-water stress coefficients are determined by the available water in the root zone (Allen et al., 1998). Actual evapotranspiration can also be measured in the field using lysimeters (Goyal and Harmsen, 2014; Allen et al., 1998). For estimating evapotranspiration, one needs to monitor the weather conditions (Allen et al., 1998). The weather stations used for measuring the weather parameters should be parameterized according to the FAO recommendation, which includes installation in an open area, surrounded by short perennial grasses not lacking in water and nutrients (Allen et al., 1998). The meteorological conditions should be monitored with sensors installed at 2 m height (Allen et al., 1998). For missing weather data, the FAO Penman-Monteith can be replaced by the Hargreaves and Samani evapotranspiration model, equation 2-24 (Bos et al., 2009; Allen et al., 1998).

43

State of the art

2.1.5.6 Effects of water stress on crops Water stress refers to the soil moisture in which the crops have low capacity to extract it (Allen et al., 1998). When the soil water content is high, the crops can transport the moisture to the atmosphere. However, due to adsorptive and capillary forces during the soil drying process, the roots become ineffective at extracting the moisture after a threshold moisture limit (Allen et al., 1998). During this time, the crops are assumed to be water stressed (Allen et al., 1998). Another concept associated with water stress is drought. Drought refers to the time when the water available in the soil is not sufficient to supply the atmospheric demand of evapotranspiration (Ehlers and Goss, 2016). Drought also indicates a period without precipitation in which the moisture in the soil is reduced and crops suffer from water deficiency (Larcher, 2003). Water deficiency in crops can also occur due to shallow soil, high evaporation, osmotic pressure, frozen soils (Larcher, 2003), or climate factors such as air humidity and temperature (He et al., 2017). The shoots and the roots of the crops identify water stress differently (Ehlers and Goss, 2016). On the shoots the guard cells lose turgor and close the stomata (Ehlers and Goss, 2016). Whereas the roots reduce the water potential and water content, decrease the contact with the soil and the water and nutrient uptake (Ehlers and Goss, 2016). Additionally, the roots will be more sensible to the mechanical impedance of the soil (Ehlers and Goss, 2016). Crops have different strategies of overcoming water deficiency and tissue desiccation during droughts (Ehlers and Goss, 2016). These strategies include drought escape, drought tolerance and drought avoidance (Ehlers and Goss, 2016). Drought escape strategies includes the early seeding dormancy and seed germination after abundant rainfall (Ehlers and Goss, 2016). Crops with drought tolerance can maintain their metabolic activities under low water potential caused by limited moisture availability (Kørup et al., 2017) and therefore show higher water use efficiency (Kørup et al., 2017). Crops also avoid droughts by reducing the stomatal conductance and increasing root to shoot ratio (Vries et al., 2016). In sloped areas, the increase in root to shoot ratio may increase the soil stability due to greater root density (Leitinger et al., 2015). Drought avoidance include crops that avoid or retard tissue desiccation by increasing or maintaining water uptake (water spenders), reducing water loss (water savers), or by storing water in their tissues (Ehlers and Goss, 2016). Crop water stress can be evaluated by observing the degree of wilting or by measuring the total plant water potential of the leaves (Ehlers and Goss, 2016). The total plant water potential is based on the principle that during water stress the water potential decreases (become more negative) because the crops lose water content from the leaves and shots (Ehlers and Goss, 2016). The leaf water pressure potential is measured using pressure chambers (Ehlers and Goss, 2016). For this a young leaf from the top of a crop shoot is wrapped with foil, and the stalk is fixed in the chamber apparatus (Ehlers and Goss, 2016). Then the xylem sap starts to flow and the pressure is recorded (Ehlers and Goss, 2016). The negative values range between the xylem and mesophyll water 44

State of the art pressure (Ehlers and Goss, 2016). The water potential is highest in the early morning and lowest in the early afternoon (Ehlers and Goss, 2016). Therefore, the measurements should be performed in the morning to verify if the water potential of the crops equilibrates during the night by the soil moisture (Ehlers and Goss, 2016). The crop water stress can also be evaluated by estimating the relative water content of the tissue using the following expression: 퐹푊 − 퐷푊 (2-77) 푅푊퐶 = 푇푊 − 퐷푊 Where FW is the fresh weight of the plant or part of the plant, DW is the dry weight, TW is the weight of the water saturated crop after being immersed overnight in water (Ehlers and Goss, 2016). Non-stressed crops have a relative water content higher than 0.8, stressed crops a relative water content between 0.72 and 0.88 and crops die when the relative water content ranges from 0.5 to 0.6 (Ehlers and Goss, 2016). Leaf temperature also indicates the crop water stress (Ehlers and Goss, 2016). Leaves have a lower temperature than the surrounding air when the transpiration meets the atmospheric moisture demand, however, the temperature difference between the leaves and the overlying air decreases when the soil moisture is reduced and the stomata closes (Ehlers and Goss, 2016). The leaves’ temperature is measured using infrared thermometers (Ehlers and Goss, 2016). Three different groups of crops can be defined according to their water relations (Kirkham, 2014), hydrophytes, mesophytes and xerophytes (Chang, 2009; Ehlers and Goss, 2016). Hydrophytes (or hygrophytes) are crops that grow in water or need a large water supply for growth (Kirkham, 2014), such as mangroves and rice (Chang, 2009). Mesophytes grow with steady available soil moisture and high relative air humidity (Kirkham, 2014). Mesophytes wilt permanently after losing 25 to 50 % of their water content (Chang, 2009). Most agricultural crops are mesophytes (Chang, 2009). Xerophytes are crops that tolerate dry environments (Kirkham, 2014). Xerophytes wilt permanently after losing from 50 to 75 % of their water content (Chang, 2009). Examples of xerophytes are desert crops (El-Keblawy et al., 2015). Larcher (2003) classify the crops as poikilohydric and homoiohydric according to their ability to compensate for changes in water supply and evaporation. The poikilohydric plants lack protection mechanisms for desiccation (Larcher, 2003). Examples of poikilohydric plants are fungi, lichens and some algae (Larcher, 2003). These organisms shrink and the metabolic activities decrease during water shortages or droughts, however, the plants recover their metabolic functions when water is available (Larcher, 2003). Metabolic activities of poikilohydric plants take place under -5 to -30 MPa, and this pressure is regulated according to the air moisture (Larcher, 2003; Toldi et al., 2009). Homoiohydric crops developed the ability to regulate the evaporation through the stomata and the protection of cuticle (Larcher, 2003). Most of the vascular and cultivated plants are sensitive to water desiccation (Toldi et al., 2009). Desiccation sensitive plants do not endure relative water content below 20 to 50 % (Toldi et al., 2009). Salt tolerant crops are termed 45

State of the art halophytes and contrast the glycophite plants which grow in soils with low sodium content (Cheeseman, 2014). Saline soils cover circa 6 % of the continental surface (Larcher, 2003). Water is essential for crop growth (Chang, 2009). Water is (1) the major constituent of the plant tissue (Chang, 2009). The water content of fresh fruits, leaves and roots ranges from 70 to 95 % (Larcher, 2003), whereas the seeds range from 5 to 15 % of water content (Larcher, 2003; Ehlers and Goss, 2016); (2) water is a reagent in the photosynthesis process, Eq. 2-78 (Chang, 2009; Taiz et al., 2015), (3) water dissolves and transports salts and sugars through the xylem and phloem, (4) water is essential for the hydrostatic pressure of plant cells, (5) and is needed for transpiration which is associated with photosynthesis and dry matter production (Chang, 2009). 6 퐶푂 + 퐻 푂 → 퐶 퐻 푂 + 6 푂 (2-78) 2 2 6 12 6 2 퐶푎푟푏표푛 푑𝑖표푥𝑖푑푒 푊푎푡푒푟 퐶푎푟푏표ℎ푦푑푟푎푡푒 푂푥𝑖푔푒푛 Water deficiency affects the growth pattern of crops, photosynthesis, transpiration rate and dry matter production (Chang, 2009). The root system is more abundant and longer under water deficit (Chang, 2009). In contrast, under frequent irrigations the root system develops horizontally and near the soil surface (Chang, 2009). Water deficit also changes the pattern of the leaves (Chang, 2009). The leaf area decreases, the thickness of the leaves increases and the ratio of root to shoot increase (Chang, 2009). The lack of moisture reduces the capacity of the crops to photosynthesize (Chang, 2009). A decrease of 30 % in the photosynthesis coincides when crops lose 30 % of the leaves water content (Chang, 2009). When the crops lose 60 % of the water content photosynthesis is interrupted (Chang, 2009). Water deficit also decreases the hydrostatic pressure of the guard cells making the crops close stomata (Chang, 2009). By closing the stomata there is a reduction in the gas exchange, transpiration and CO2 uptake (Chang, 2009). By reducing the transpiration, the crops are unable to reduce the leaf and air temperature by water evaporation or translocate nutrients from the roots (Chang, 2009). For actively growing crops transpiration is associated with dry matter production (Chang, 2009; Jensen et al., 2001). Jensen et al. (2001) evaluated the dry matter production of different grasses under 5 water irrigation levels in the United States during 1997 and 1998. The water levels were on average 91, 78, 63, 55, 41 cm/year. The dry matter production of perennial ryegrass-hybrids (Lolium perenne L.) were 11.1, 10.8, 10.2, 8.8 and 6.4 tons/ha, respectively. On average, the highest dry matter, 20.9 ton/ha, was found in tall fescue (Festuca arundinacea Schreb.) and the lowest in the perennial ryegrass, 9.5 ton/ha. The authors found a mostly linear association between the dry matter and the water levels, except for lower water levels which had a quadratic component. He et al. (2017) evaluated the biomass production of ryegrass during a drought and rehydration period in New Zealand (He et al., 2017). The drought period occurred after crop establishment from 20 December 2012 to 15 March 2013. The rehydration period was induced during the two subsequent months (April and May). The total drought period was 85 days. The treatment 46

State of the art comprised irrigated and non-irrigated crops (He et al., 2017). During the drought period the crops received no irrigation (He et al., 2017). The authors found the drought decreased the dry matter of the crops which is associated with a decrease in plant size, leaf elongation and accelerated leaf senescence (He et al., 2017). This decreased the water demand of the crops (He et al., 2017). After drought, during the rehydration months in autumn, one found the non-irrigated crops presented similar shoot biomasses as the irrigated treatments (He et al., 2017). However, the authors highlighted that the biomass production after the drought can be associated with the growth stage during the drought (He et al., 2017). When the water deficit occurs during vegetative growth, the crops present faster regrowth than when the water stress occurs during flowering (He et al., 2017). The rapid regrowth of the crops can be associated with the available nutrients in the soil from the drying period (He et al., 2017). Hendrickson et al. (2013) analyzed the water use efficiency of switchgrass (Panicum virgatum L.), western wheatgrass (Pascopyrum smithii (Rydb.) Á. Löve), and a western wheatgrass–alfalfa (Medicago sativa L.) mixture under two induced droughts during the growing season. The experiment had a control treatment which received the precipitation from May to August and a non- irrigated treatment which received 50 % of the control precipitation (Hendrickson et al., 2013). The precipitation in the drought treatments was divided into two periods. During early water stress, the crops received 20 % of the precipitation in May and June and 80 % in July and August. During the late water stress, the crops received 80 % in May and June and 20 % in July and August (Hendrickson et al., 2013). The drought was induced using a rainout shelter (Hendrickson et al., 2013). The study was carried out in 2006 and in 2007 (Hendrickson et al., 2013). The water use efficiency was estimated using the expression: 푏𝑖표푚푎푠푠 (2-79) 푊푈퐸 = (퐼푤푎푡푒푟 − 퐹푤푎푡푒푟) + 퐼푅푅

Where WUE is the water use efficiency (g/mm), Iwater is the soil moisture before the first irrigation (mm), Fwater is the final water content after the killing frost (mm), IRR is the May - August irrigation (mm), biomass is the dry matter yield (g) at the end of season obtained by ovendrying the crops under 55 oC until constant weight (Hendrickson et al., 2013). The authors found no significant differences among the crops for the water use efficiency in the control treatments, however switchgrass presented higher water use efficiency under the control treatment, 5.64 g/mm, and drought treatments, 6.41 g/mm under the early drought and 7.42 g/mm under the late drought. Water use efficiency of perennial grasses for bioenergy production shows the capacity of the crops to provide biomass under different precipitation levels and water availability (Hendrickson et al., 2013). Leitinger et al. (2015) studied the impacts of drought on the runoff and soil water content on two sub-watersheds located in the French and Austrian Alps. Runoff from mountain areas is important 47

State of the art because more than 50 % of the world´s population rely on water provision from mountains (Leitinger et al. (2015). This runoff is used for forage production, hydropower and grazing (Leitinger et al. (2015). The authors simulated the runoff using the physically based Hillflow model. The drought was simulated by reducing the long-term historical average precipitation from 1970 to 2000 to the levels of the very dry year of 2003 during the spring and summer months. The normal precipitation in the French Alps was 355.1 mm and under dry conditions, 152.3 mm, whereas in the Austrian Alps these levels were 631.6 mm and 227.3 mm respectively. The study was performed from May to September. The authors also simulated the moisture content at the main root depth from 0.0 to 0.3 m depth. Leitinger et al. (2015) found a decrease trend for the soil moisture content and runoff due to droughts. However, the decrease was less pronounced in the drier French Alps and this shows the water saving strategy of the crops. In the Austrian Alps, known to be more humid, the reduction in water content and seepage was more pronounced and the crops were considered as water-spending. In addition, in the French Alps, the reduction in moisture content was more pronounced in intensive managed crop systems (fertilized, mowed and terraced). However, no differences were found among the crop management systems in the Austrian Alps. The authors highlight that the crops’ water-saving strategy may develop in plants exposed to frequent dry environments. Under a normal year, the evapotranspiration precipitation ratios were 0.66 for the French Alps and 0.58 for the Austrian Alps. Whereas under dry conditions these ratios were 0.80 and 0.89 respectively (Leitinger et al., 2015). Kørup et al. (2017) evaluated the drought tolerance of six perennial grasses, including four C3 and two C4 species. For this Kørup et al. (2017) evaluated the aboveground biomass, the stomatal conductance, leaf water potential, water use efficiency and regrowth after drought. The study was performed using 1 m² drainage lysimeters at a depth of 1.4 m. The water stress was applied by drying the soil up to 80 % of the root zone water available capacity (Kørup et al., 2017). The study was performed in Denmark from May to October of 2014 and 2015. The treatments consisted of irrigated and rainfed conditions. The authors observed that the dry matter yield decreased during the drought, however after the water stress, the crops exposed to drought had higher yields than the fully irrigated treatment. Moreover, the water use efficiency was higher and the water potential and stomatal conductance were lower during the drought treatments when compared with the control.

48

State of the art

Water stress on evapotranspiration studies Water stress can be considered in non-standard evapotranspiration conditions using a water stress coefficient (Allen et al., 1998), as follow:

퐸푇푐 푎푑푗 = (퐾푠퐾푐푏 + 퐾푒) 퐸푇0 (2-80)

퐸푇푐 푎푑푗 = 퐾푠 퐾푐 퐸푇0 (2-81)

Where ETc adj is the adjusted evapotranspiration (mm), ET0 is the reference evapotranspiration

(mm), Kcb represents the transpiration component of the evapotranspiration and Ke refers to the evaporation component of the adjusted evapotranspiration. For water stress conditions, Ks <1, for non-water stress conditions, Ks = 1 (Allen et al., 1998). When using the single crop coefficient, the water stress coefficient adjusts the crop coefficient, Eq. 2-81 (Allen et al., 1998). The water stress coefficient can be estimated using the expression (Allen et al., 1998): 푇퐴푊 − 퐷 푇퐴푊 − 퐷 (2-82) 퐾 = 푟 = 푟 푠 푇퐴푊 − 푅퐴푊 (1 − 푝) 푇퐴푊

Where Ks is the transpiration reduction dependent on available soil moisture (0-1), Dr is the root zone depletion (mm), TAW is the total available soil moisture in the root zone (mm), p is the fraction of TAW that a crop can extract from the root zone without suffering water stress (-) (Allen et al., 1998). The p value is available for several crops in official publications (Allen et al., 1998) or obtained experimentally (Bilibio et al., 2014). The minimum root zone moisture depletion value is registered at the field capacity and the maximum value is equal to the total available water (Allen et al., 1998). The moisture content above field capacity is considered as drainage and the moisture content below the permanent wilting point is unavailable for crop extraction (Allen et al., 1998). The total water available is estimated using the expression:

푇퐴푊 = 1000 (휃퐹퐶 − 휃푊푃)푍푟 (2-83)

Where TAW is the total available soil moisture in the root zone (mm), θFC is the water content at field capacity (m³/m³), θWP is the water content at the permanent wilting point (m³/m³), Zr is the root depth (m) (Allen et al., 1998). The ready water available can be written as:

푅퐴푊 = 푝 푇퐴푊 (2-84) Where RAW is the ready available water (mm), p is the fraction of the total available water that can be extracted by the roots without crop water stress or reduction in evapotranspiration. The p values range from 0.3 for crops with shallow roots to 0.7 for deep-rooted crops (Allen et al., 1998). A value of 0.5 is generally used (Allen et al., 1998).

49

State of the art

2.2 Mining and potash tailings Mining is one of the first economic activities and up to this day still supplies many basic resources for living (Hartman and Mutmansky, 2002). Mining is the extraction of any solid, liquid or gas from the earth for utilitarian purposes (Hartman and Mutmansky, 2002). The minerals mined are classified into metallic (i.e. ores which contain iron, copper, nickel, lead, zinc, aluminium, gold, etc.); non-metalic (i.e. salt, clays, sand, gravels, fertilizers); and fossil fuels (coal, oil, natural gas) (Hartman and Mutmansky, 2002). There are surface and underground methods for mineral extraction. On the one hand, surface mining is used to remove minerals near the ground using mechanical excavation, such as open-pit and open-cast methods (Hartman and Mutmansky, 2002). On the other hand, underground mining allows miners to descend below the surface to remove mineral ores located in deeper layers. Currently, circa 60 % of mining activities are surface based (Hartman and Mutmansky, 2002). Mining is closely related to agriculture, providing fertilizers for crop growth and animal feeding, as well as metals and fuel for machinery. Mining activities also contribute to economic development, the creation of wealth (Paredes, 2016), jobs (Wessman et al., 2014) and infrastructure (Hartman & Mutmansky, 2002). In Germany, circa 800 million tones are mined annually (Wedig, 2014). Gravel and sand mining make up 30 % of this amount (Wedig, 2014). Lignite, hard coal, natural gas and crude oil account for another 209.6 million tons or 26.2 % of the total mined products (Wedig, 2014). The potash magnesium mining accounts for 3 % of Germany’s annual mining volume (Wedig, 2014). In 2013, potash mining made circa 6 million tons of K2O equivalent (Perez, 2016). Germany was the first country to perform subsurface potash mining (Warren, 2016). The potash mining activities in Germany started in Staßfurt, state of Sachsen-Anhalt, in 1861 (Warren, 2016; Ciceri et al., 2015). Up to that date, potassium compounds in Europe were obtained from the evaporation of leached ashes (Ciceri et al., 2015). These ashes originated from trees in Russia, halophyte plants (i.e., Salsola soda L.) from Mediterranean countries, such as Italy and Spain, and marine algae from northern European countries (Ciceri et al., 2015). According to Larcher (2003), crops that grow under saline soils have higher rates of ashes or mineral content. Seaweeds and halophytes have from 10-20 % of ashes in the dry matter, whereas the ash content of the leaves from woody plants varies from 4 to 9 % (Larcher, 2003). The largest known potash reserves are concentrated in a few regions of the globe (al Rawashdeh and Maxwell, 2014). Canada has 45 % of the world´s known potash reserves, Russia 35 % and Belarus 9 % (Warren, 2016). In Europe, the potash reserves are in Germany (Zechstein basin), Spain, United Kingdom and in the Netherlands (Warren, 2016). The top ten potash producers are Russia, Canada, Belarus, China, Germany, Israel, Jordan, Chile, The United States and United Kingdom, based on the year 2013 (Warren, 2016). 50

State of the art

Beyond the concentration of potash reserves, potash production is also controlled by a few companies (al Rawashdeh and Maxwell, 2014). The largest companies performing potash mining worldwide in a decreasing order are Belaruskali (Belarus), Potash Corp (Canada), Mosaic (Canada), ICL (Israel, Spain and Uk), Urakali (Russia), Silvinit (Russia), K+S KALI (Germany), Sinofert (China), APC (Jordan), SQM (chile), Agrium (Canada, Intrepid (USA), Vale (Brazil) (Warren, 2016). Just 10 companies control 90 % of the world´s potash and the largest 5 companies control circa 70 % of the world´s potash (Warren, 2016). The company extracting potash in Germany, K+S KALI, is in seven different regions in four different states, Lowe Saxony, Saxony-Anhalt, Hessen and Turingia (Wedig, 2014). In Hessen the Werra potash facility combines the potash processing of Wintershall, Hattorf and Unterbreizbach (Rauche, 2015). Annually, a potash-magnesium production of 3.01 million tons is expected in the Werra combined plant (Rauche, 2015). For this, an additional 14 million tons of solid wastes as well as 7.5 million cubic meters of liquid wastes are produced (Rauche, 2015). From the solids, 1.4 million tons are backfilled underground and 13 million tons are heaped above ground near the processing facilities (Rauche, 2015). Whereas 4 million m³ are pressed underground and 3 million m³ are disposed in water courses (Rauche, 2015). The liquid wastes are saline brines with on average 37 % salt content (Rauche, 2015). The solid tailings at the Werra combined plant are composed of sodium chloride, 91 %; potassium chloride, 1 %; magnesium sulfate, 5 %; calcium sulfate, 2 % (European Commission, 2009). These solid residues on the surface are exposed to precipitation, which generate additional brines (European Commission, 2009). These brines are collected in seal reservoirs and gradually disposed in surface waters (European Commission, 2009). In Germany, circa 500 ha are covered by solid potash tailings piles, such as Wintershall, 92 ha; Hattorf, 81 ha; Zielitz, 192 ha; Neuhof – Ellers, 83 ha; Sigmundshall, 50 ha (Rauche, 2015). On the pile, the tailings harden and crystalize after a short period of time (European Commission, 2009). The bulk density of the piles at the impermeable core of the heaps is 2.2 g/cm³, which is similar to the original underground density (European Commission, 2009). The impermeable core of the heap is located between 20 and 80 meters below the pile´s surface (European Commission, 2009). The disposal of brines in water courses and underground in the Werra region has raised several environmental issues with the water supply sector and local communities (Stürmer, 2012; Winter, 2016; Braukmann and Böhme, 2011). Official regulation has been imposed thorough the time to limit the rate of chloride content (Braukmann and Böhme, 2011). Currently the upper limit of salt is 2.5 mg/liter at the Gerstungen site of the Werra River (Braukmann and Böhme, 2011; Coring and Bäthe, 2011). In addition, the European Water Framework Directive (EU WFD, 2000) demands that all European rivers present good ecological and chemical statuses by 2027 (European Parliament, 2000; European Comission, 2012; Braukmann and Böhme, 2011).

51

State of the art

Considering this, several studies have been performed to evaluate the use of static and non-static covers of potash tailings piles to reduce the precipitation erosion and leaching of salt, including Hermsmeyer (2001), Niessing, (2005), Papke and Schmeisky (2013), Scheer (2001), Schmeisky and Hofmann (2000). However, no study considered the use of technogenic substrates made of household waste incineration slags and coal combustion residues as non-static covers on potash tailings piles.

2.3 Evapotranspiration covers and technogenic substrates Evapotranspiration covers, also known as water balance covers, soil-plant covers or store-and- release covers (Rock, et al., 2012), use a precipitation soil reservoir and a vegetation cover to move the moisture back to the atmosphere (Hauser, 2009). In evapotranspiration covers, the water from precipitation infiltrates the soil and fills the pore space (Hauser, 2009). At field capacity or below it the moisture retained flows downwards at a very low rate in the soil profile (Hauser, 2009). Whereas the evaporation from the soil and the transpiration of the crops increase the hydraulic gradients and reverse water movement to the surface (Hauser, 2009). The evapotranspiration on the surface is enhanced by the vegetation roots, removing the moisture from deeper layers (Hauser, 2009). These evapotranspiration covers’ principles have been common knowledge for a long time, however their application in waste systems is more recent (Hauser, 2009; Rock, et al., 2012). The concept of evapotranspiration landfill covers was first published in 1994 (Hauser et al., 2005). Evapotranspiration covers are alternative to conventional capillary barriers in waste systems (Zhang and Sun, 2014). Capillary barriers rely on differences in hydraulic conductivity of the cover materials to minimize the water percolation into the wastes and increase the runoff (Rock, et al., 2012; Zhang and Sun, 2014). The problems related with capillary barriers are installation costs and preferential flow (Rock, et al., 2012). High installation costs are associated with the need to compact the soil (Schnabel et al., 2012) and maintenance, i.e. close secondary pores and cracks which speed water infiltration (Schnabel et al., 2012). These cracks are caused by freezing and thawing cycles, as well as the drying process of the compacted materials (Schnabel et al., 2012). As evapotranspiration covers rely on natural processes (soil water storage and evapotranspiration), they are expected to function for decades or centuries and be less expensive (Hauser et al., 2001; Hauser et al., 2005). The establishment of evapotranspiration covers is 35 to 72 % of the costs to implement capillary barriers (Hauser, 2009). The costs to implement a capillary barrier in the United States ranges from $240,000 to $890,000/ha and to implement evapotranspiration covers the costs vary from $200,000 to $740,000/ha (Barnswell and Dwyer, 2012). The design of evapotranspiration covers should consider historical weather conditions, soil type and thickness, vegetation types and soil fertility (Rock, et al., 2012). The most important weather parameter is the historical averages, annual distribution and extreme precipitation events (Hauser, 52

State of the art

2009). The precipitation levels help to define the thickness of the monolithic covers (Rock, et al., 2012). Air temperature, solar radiation, air humidity and wind speed also determine the suitability of evapotranspiration covers (Hauser, 2009; Schnabel et al. 2012; Hauser et al., 2001). Evapotranspiration covers are more favorable in arid and semi-arid climates because the ratio of evaporation to precipitation is more favorable (Hauser et al., 2001). However, in humid and temperate environments, the efficiency of evapotranspiration covers may be reduced due to higher precipitation levels and lower evapotranspiration capacities (Rock, et al., 2012). Arid and semiarid regions have a precipitation to potential evapotranspiration ratio lower than 0.75 whereas humid areas have a precipitation to potential evapotranspiration ratio higher than 0.75 (Barnswell and Dwyer, 2012). Regarding the soil, Hauser (2009) suggests the use of local materials to reduce transport costs. The soil should have a high water retention capacity and fertility to guarantee the vegetation’s growth (Hauser, 2009). The thickness of the soil should be studied using mathematical models and extreme precipitation events (Hauser, 2009). The soil thickness may range from 0.2 to 2.0 m in common covering systems (Hauser et al., 2001). The vegetation covers should include a mixture of native perennial grasses (Hauser, 2009). Perennial grasses avoid the need to reseed, have different growth stages and drought tolerances (Hauser, 2009). Native crops have already endured drought periods or even former fires (Hauser et al., 2001). Evapotranspiration covers are used to reduce the water infiltration into waste systems, decrease the seepage from the wastes to ground water, control emission of gases (Hauser et al., 2001), decrease erosion and exposure to the wastes and restore the landscape (Rock et al., 2012; Hauser, 2009). Evapotranspiration covers can be used in municipal landfills, industrial landfill or mine sites (Hauser et al., 2001). The performance of evapotranspiration covers is generally evaluated by measuring or estimating the seepage (Rock, et al., 2012). This is achieved by determining all water balance components (Hauser, et al., 2005). Estimating evapotranspiration is important because it is the largest output term and controls the size of the other water balance components (Hauser et al., 2005). Soil substitutes can also be used as single layers to cover waste systems and support vegetation growth (Hauser, 2009; Rock, et al., 2012). These soil substitutes are termed technosols (Howard, 2017), technogenic substrates (Blume et al., 2016), artificial or man-made substrates (Meuser, 2012). Technosols are soils formed from human transported materials, such as artifacts. The rate of artifacts in technosols is equal or larger than 20 % by volume in the upper 100 cm (Howard, 2017). These artifacts are waste building materials such as brick, mortar, industrial wastes such as slags, or ceramic objects (Howard, 2017). Artifacts have diameters larger than 2 mm and microartifacts are 0.25 - 2.0 mm in size (Howard, 2017). Examples of techcnosols are soils containing wastes from landfills and coal ash (Howard, 2017).

53

State of the art

Research has been conducted to evaluate the use of evapotranspiration covers and technogenic substrates. Schnabel et al. (2012) evaluated the use of two different landfill covers, a compacted soil cover (CSC) and an alternative evapotranspiration cover (ET cover), in Southcentral Alaska. The authors used drainage lysimeters to measure the seepage amount (Schnabel et al., 2012). The mean temperature of the experimental site was 2 oC and the annual average precipitation was 406 mm. The reference evapotranspiration ranged from 480 mm to 670 mm/year. The experiment was performed from 1 August 2005 to 31 July 2009 (Schnabel et al., 2012). After four years of evaluation the lysimeters received 1636 mm of precipitation (Schnabel et al., 2012). The seepage amount was 201 mm in the evapotranspiration cover and 292 mm for the compacted soil (Schnabel et al., 2012). The actual evapotranspiration was 1360 mm (83 %) in the ET lysimeter and 1236 mm (76 %) in the CSC lysimeter (Schnabel et al., 2012). The cumulative drainage of the evapotranspiration cover started to diverge from the compacted soil cover after the second year (Schnabel et al., 2012). An increase of the performance is expected from evapotranspiration covers with time because the crops are better established (Schnabel et al., 2012). Whereas the performance of compacted soil covers decrease with time due to the formation of preferential flow (Schnabel et al., 2012). Barnswell and Dwyer (2012) studied the percolation rate of an evapotranspiration cover in Oshio, USA. The authors used drainage lysimeters and two different crops, mature plants and immature plants. The mature plants were transferred from a 10-year tall-grass field and the immature plants were seeded after lysimeter installation. The author used dredged sediments mixed with organic materials as cover soil. The experiment was carried out over two years, from 10 June 2009 to 10 June 2011. The regional air temperature ranged from 27.4 oC in the summer to -7.1 oC in the winter, whereas the annual mean precipitation is 83.5 cm and the reference evapotranspiration, 64.7 cm. Barnswell and Dwyer (2012) applied a precipitation of 94 cm in the first year and 69 cm in the second year. In the first year the mature crops produced less seepage than the immature mixture crops. In contrast, in the second year the immature crops showed lower seepages than the mature crops. The seepage in the first year was 4 cm for the mature crops and 17 cm for the immature crops. In the second year the mature crop seepage was 10 cm and the immature crops 3 cm of seepage. These differences are associated with the biomass increase of the immature plants in the second year (Barnswell and Dwyer, 2012). Nyhan (2005) evaluated the evaporation, runoff, and seepage of an unvegetated evapotranspiration cover under 5, 10, 15 and 25 % of slope exposed to field precipitation. The evapotranspiration cover was made of 15 cm loam soil, 76 cm of crushed turf and 30 cm medium gravel. The water soil erosion on the top was minimized using medium sized gravel. The study was carried out in the semiarid-temperate region of New Mexico from 1992 to 1998, totaling 7 years (Nyhan, 2005). The average precipitation of the experimental region from 1911 to 1986 was 46.9 54

State of the art cm/ year, almost 50 % of the precipitation was registered from July to August. Nyhan (2005) found an evaporation rate ranging from 88 to 95 %, a runoff of up to 4 % and a maximum seepage of 1.7 %. The increasing slope decreased the seepage of the evapotranspiration cover. In contrast, the increasing slope increased the runoff and the evaporation rate of the soil cover. The increase in evaporation in slope areas was due to the high interception of solar radiation (Nyhan, 2005). The authors report that a vegetation surface would minimize the runoff. However, the contribution of the transpiration from the crops to the precipitation losses would have been minimal due to severe droughts in the region (Nyhan, 2005). Wang (2017) studied the use of green roofs in urban areas. The aim of the green roof is to store the water in the vegetated soil or substrates. The author suggested that using medium crop status and low fertilization levels reduces the chemical composition of the runoff. Additionally, the increased depth of the roofs increases the water storage capacity but may also increase the rate of pollutants in the outflow. Asensio et al. (2013) studied the use of technosols made of organic wastes, sewage sludge and paper mill ashes as soil amendments in a copper mine tailings pile in Spain. The technosols were distributed over the mined soil making up a 5-cm layer. The soil amendment improved the organic carbon and pH of the mined soils. However, the technosols increased the rate of Nickel, Zinc and Lead in the mined soil. Therefore the authors recommended verifying the rate of chemical elements in the technosols before implementation.

2.3.1 Perennial grasses Perennial grasses, such as perennial ryegrass, red fescue and Kentucky bluegrass, are crops that last longer than two years (Darke, 1994). Perennial ryegrass has rhizomes distributed on the subsurface and tufted stems (Darke, 1994). Rhizome are specialized stems located on the soil subsurface producing roots, leaves and inflorescences (Darke, 1994). The leaves are circa 30 cm long (Darke, 1994). Ryegrass is often used as pasture and turf (Darke, 1994). Perennial ryegrass (Lolium perenne L.) is widely distributed in temperate regions (He et al., 2017). In Western Europe the growing season of perennial ryegrass is from March to November (Boller et al., 2010). Perennial ryegrass is used for pasture in Europe, New Zealand, northwestern USA and Canada (Jensen et al., 2001). However, the density of the perennial ryegrasses decreases quickly (Jensen et al., 2001), hence reseeding may be recommended (Jensen et al., 2001). The potential yield of perennial ryegrass under 1332 mm of precipitation in New Zealand ranges from 15.8 tons of dry matter/hectare in cooler conditions to 19.9 tons of dry matter/hectare in warmer conditions (He et al., 2017). The dry matter production of ryegrass in Utah, United States, under optimum precipitation and irrigation conditions (91 cm) was 11.1 t ton/ha/year and under restricted water use (41 cm), the dry matter yield was 6.4 ton/ha/year (Jensen et al., 2001). This yield was obtained by

55

State of the art harvesting the grasses at a height of 30 cm circa 5 times from May to October of 1997 and 1998 (Jensen et al., 2001). Red fescue has stems of up to 110 cm, is loosely tufted, tolerates up to -15 oC and present long rhizomes (Darke, 1994). Fescues tolerate droughts due to the very thin leaves which minimize transpiration (Boller et al., 2010). Fescues tolerate low fertile soils and low soil pH levels (Boller et al., 2010). Red fescue is used for forage, turf, landscape and ornamental purposes (Boller et al., 2010). Kentucky bluegrass has stems of up to 90 cm, is adapted to cool weather, and is loosely tufted (Darke, 1994). Kentucky bluegrass is adapted to temperate climates and is used for pasture and turf (Boller et al., 2010). Kentucky bluegrass has a low tolerance to drought, heat and soil salinity (Boller et al., 2010). In evapotranspiration covers, a mixture of perennial grasses is recommended (Hauser, 2009). Kørup et al. (2017) highlight that rhizomatous grasses often have high tolerances to drought and yield potential.

2.3.2 Waste system in Germany Waste refers to all substances or objects that must be discarded by the holders (BMUB, 2012). In Germany, the total amount of waste was 339.132 million tons in 2013 (Nelles et al., 2016). The highest amount of waste originated from construction and demolition (202.735 millions of tons) and the lowest from mining rubble (29.250 millions of tons) (Nelles et al., 2016). The recycling rate was highest in the construction and demolition waste and in the municipal solid waste (87 %) and lowest in the mining industry (1 %) (Nelles et al., 2016). Since June 2005 the municipal solid wastes that are not recycled, must receive biological or thermal treatment in Germany prior to being landfilled (Nelles et al., 2016). Moreover, the energy generated by the treatment must be utilized (Nelles et al., 2016). The thermal treatments of the municipal solid wastes are performed to eliminate microorganisms, to concentrate inorganic constituents, reduce organic matter and reduce the total volume and mass of the residues (Sabbas et al., 2003). The incineration processes reduce the initial volume by 90 % (Sabbas et al., 2003). The main products generated by the incineration of municipal solid wastes are bottom ashes, boiler ash and air control residues (Sabbas et al., 2003). The largest portion of the residues is bottom ashes, up to 80 % of the residues (Inkaew et al., 2016), which is obtained from the burning process (Sabbas et al., 2003). The boiler ash originates from a heat recovery system and air control residues are recovered before the injection of air into the atmosphere (Sabbas et al., 2003). The amount of bottom ashes corresponds to 250 kg/ton of municipal solid wastes (Holm and Simon, 2017). These residues are used in construction, roads or landfilled (Holm and Simon, 2017). Annually, 5 million tons of bottom ashes are generated in Germany (Holm and Simon, 2017). The composition of the bottom ashes depends on the initial composition of the wastes (Holm and Simon, 2017). Although a general

56

State of the art composition of bottom ashes includes 1 % unburned materials, 9 % glass cullet, 10 % metals, 40 % ashes and 40 % melting products (Holm and Simon, 2017). Wastes in Germany are also produced by coal combustion. The combustion of coal provides 45 % of the electricity in Germany (Heinrichs et al., 2017). Several residues are produced during coal combustion, such as fly ashes, bottom ashes, boiler slag and flue-gas desulfurization by-products (Park et al., 2014). Annually circa 22 million tons of coal combustion residues are produced in Germany (Feuerborn et al., 2012). These materials are used mainly in construction and refilling mining voids (Feuerborn et al., 2012).

2.3.3 Mathematical models for evapotranspiration covers A model is a representation of a system (Tucci, 2005). A system is any structure of the reality which responds to information or energy (Tucci, 2005; Lascano, 1991). Whereas simulation refers to the building and adjustments of models to study a system (Lascano, 1991). Several models have been used to facilitate the understanding of crop response to the environment (Jones et al., 2017). The CropWat (Smith, 1992; FAO, 2014) and Hydrus (Šimůnek et al., 2008) have been widely used. CropWat is a computer program to calculate reference evapotranspiration, crop water requirements, irrigation requirements and the water supply scheme based on soil, climate and crop data (Smith, 1992; Stancalie et al., 2010). Standard crop data are included in the program and climatic data can be obtained for 144 countries through the ClimWat- database (FAO, 2014). CropWat version 5.7 was developed by the Land and Water Development Division of the Food and Agriculture Organization of the United Nations (FAO) and issued in 1992 (Smith, 1992). Currently CropWat version 8.0 can be downloaded from FAO's server (FAO, 2014). The main inputs are monthly climatic data (rainfall, minimum and maximum temperature, relative air humidity, sunshine duration, wind speed), cropping pattern (planting date, crop coefficient, crop description, maximum rooting depth), soil type (initial soil moisture condition, maximum rain infiltration rate) and scheduling criteria (Stancalie et al., 2010). Hydrus is one of the most robust mathematical models for simulating water flow up to three directions, x, y and z (Šimůnek et al., 2008). Hydrus simulates infiltration, evaporation, transpiration, redistribution and discharge water through saturated and unsaturated mediums using Richards equation (Šimůnek et al., 2008; Radcliffe and Simunek, 2010; Šimůnek et al., 2013). Heat and solute that flow within the medium can also be studied with advection-dispersion equations (Radcliffe and Simunek, 2010; Lamb, 2015). Models are useful tools to evaluate the performance of evapotranspiration covers because it is not possible to test ET covers in every landfill site (Hauser, 2001) and climatic regions (Schnabel et al., 2012). Model simulations of evapotranspiration covers should consider different soil types, plant cover, soil tickness and weather conditions (Hauser et al., 2001). Moreover, historical series 57

State of the art of weather data should be considered to provide the responses (i.e., seepage) of evapotranspiration covers for extreme weather events (Hauser et al., 2001; Hauser et al., 2005).

2.4 References Aboukhaled, A., Alfaro, A., Smith, M., 1982. Lysimeters, FAO Irrigation and Drainage Paper 39, Rome. Abtew, W., 1996. Evapotranspiration measurements and modeling for three wetland systems in south Florida. J. Am. Water Resour. Assoc. 32, 465-473. Doi. 10.1111/j.1752- 1688.1996.tb04044.x. Abtew, W., Melesse, A.M., 2013. Evaporation and evapotranspiration. Measurements and estimations. London: Springer. al Rawashdeh, R., Maxwell, P., 2014. Analyzing the world potash industry. Res. Policy 41, 143- 151. Doi: 10.1016/j.resourpol.2014.05.004. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration - Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, Rome. Allen, R.G., Pruitt, W.O., 1986. Rational use of the FAO Blaney‐Criddle formula. J. Irrig. Drain. Div. 112, 139-155. Doi: 10.1061/(ASCE)0733-9437(1986)112:2(139). Asensio, V., Vega, F.A., Andrade, M.L., Covelo, E.F., 2013. Technosols made of wastes to improve physico-chemical characteristics of a copper mine soil. Pedosphere 23, 1-9. Doi: 10.1016/S1002- 0160(12)60074-5. Bajraktari, N., Hélix-Nielsen, C., Madsen, H.T., 2017. Pressure retarded osmosis from hypersaline sources - A review. Desalination 413, 65-85. Doi: 10.1016/j.desal.2017.02.017. Barnswell, K.D., Dwyer, D.F., 2012. Two-year performance by evapotranspiration covers for municipal solid waste landfills in northwest Ohio. Waste Manage. 32, 2336-2341. Doi: 10.1016/j.wasman.2012.07.014. Bernardo, S., Soares, A.A., Mantovani, E.C., 2006. Manual de irrigação. (8 ed.). Federal University of Viçosa, Viçosa. Bethune, M.G., Selle, B., Wang, Q.J., 2008. Understanding and predicting deep percolation under surface irrigation. Water Resour. Res. 44 W12430. Doi: 10.1029/2007WR006380. Bilibio C., Carvalho J.A., Hensel O., Fraga A.C., Richter U., Rezende F., 2014. Effects of different soil water tensions on rapeseed crops (Brassica napus L.). Agric. Eng. Int. CIGR J 16, 1-11. Blaney, H.F., Criddle, W.D., 1950. Determining water requirements in irrigated areas from climatological and irrigation data. Technical Paper 96, 1-48. United States Department of Agriculture, Soil Conservation Service, Washington - District of Columbia. Blume, H.P., Brümmer, G.W., Fleige, H., Horn, R., Kandeler, E., Kögel-Knabner, I., et al., 2016. Scheffer/Schachtschabel Soil Science. (1 ed.). Berlin: Springer. 58

State of the art

Boller, B., Posselt, U.K., Veronesi, F., 2010. Fodder crops and amenity grasses. (1 ed.). New York: Springer. Bos, M.G., Kselik, R.A.L., Allen, R.G., Molden, D.J., 2009. Water requirements for irrigation and the environment. Dordrecht: Springer. Bradford, S.A., Headd, B., Arye, G., Simunek, J., 2015. Transport of E. coli D21g with runoff water under different solution chemistry conditions and surface slopes. J. Hydrol. 525, 760-768. Doi: 10.1016/j.jhydrol.2015.04.038. Brady, N.C., Weil, R.R., 2014. The nature and properties of soils. (14 ed.). Harlow: Pearson. Braukmann, U.; Böhme, D., 2011. Salt pollution of the middle and lower sections of the river Werra (Germany) and its impact on benthic macroinvertebrates. Limnologica 41, 113-124. Doi: 10.1016/j.limno.2010.09.003. Buckingham, E., 1907. Studies on the movement of soil moisture. U.S. Department of Agriculture, Bureau of Soils, Bulletin no. 38. Burt, S., 2012. The weather observer's handbook. Cambridge: Cambridge University Press. Čadro, S., Uzunović, M., Žurovec, J., Žurovec, O., 2017. Validation and calibration of various reference evapotranspiration alternative methods under the climate conditions of Bosnia and Herzegovina. Int. Soil Water Conserv. Res. Doi: 10.1016/j.iswcr.2017.07.002. Caiqiong, Y., Jun, F., 2016. Application of HYDRUS-1D model to provide antecedent soil water contents for analysis of runoff and soil erosion from a slope on the Loess Plateau. Catena 139, 1-8. Doi: 10.1016/j.catena.2015.11.017. Chang, J., 2009. Climate and agriculture. An ecological survey. London: Aldine Transaction. Cheeseman, J.M., 2015. The evolution of halophytes, glycophytes and crops, and its implications for food security under saline conditions. The New Phytol. 206, 557-570. Doi: 10.1111/nph.13217. Ciceri, D., Manning, D.A.C., Allanore, A., 2015. Historical and technical developments of potassium resources. Sci. Total Environ. 502, 590-601. Doi: 10.1016/j.scitotenv.2014.09.013. Coring, E., Bäthe, J., 2011. Effects of reduced salt concentrations on plant communities in the River Werra (Germany). Limnol. Ecol. Manage. Inland Waters 41, 134-142. Doi: 10.1016/j.limno.2010.08.004. Darke, R., 1994. Manual of grasses. Portland: Timber Press. Doorenbos, J., Pruitt, W.O., 1977. Guidelines for predicting crop water requirements. FAO Irrigation and Drainage Paper 24, Rome. Droogers, P., Allen, R.G., 2002. Estimating reference evapotranspiration under inaccurate data conditions. Irrig. Drain. Syst. 16, 33-45. Doi: 10.1023/A:1015508322413. Edwards, P.J., Williard, K.W.J., Schoonover, J.E., 2015. Fundamentals of watershed hydrology. J. Contemp. Water Res. Educ. 154, 3-20. Doi: 10.1111/j.1936-704X.2015.03185.x. 59

State of the art

Ehlers, W., Goss, M., 2016. Water dynamics in plant production. (2. ed.). Wallingford: CABI Publishing. El-Keblawy, A., Abdelfattah, M.A., Khedr, A.A., 2015. Relationships between landforms, soil characteristics and dominant xerophytes in the hyper-arid northern United Arab Emirates. J. Arid Environ. 117, 28-36. Doi: 10.1016/j.jaridenv.2015.02.008. European Commission, 2009. Reference document on best available techniques for management of tailings and waste-rock in mining activities. http://eippcb.jrc.ec.europa.eu/reference/BREF/mmr_adopted_0109.pdf (accessed 17 August 2016). European Parliament, 2000. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000. Establishing a framework for community action in the field of water policy. http://data.europa.eu/eli/dir/2000/60/oj (accessed 05 September 2016). Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety (BMUB), 2012. Reorganising the law on closed cycle management and waste http://www.bmub.bund.de/en/topics/water-waste-soil/waste-management/waste-policy/cycle- management/artikel/law-on-closed-cycle-management/ (accessed 08 December 2017). Feuerborn, H., Müller, B., Walter, E., 2012. Use of calcareous fly ash in Germany. Proceedings of the EUROCOALASH 2012 Conference, Thessaloniki Greece. Food and Agriculture Organization (FAO), 2014. CROPWAT software. FAO Land and Water Division, Rome. http://www.fao.org/land-water/databases-and-software/cropwat/en/ (accessed 17 November 2017). Goyal, M.R., Harmsen, E.W., 2014. Evapotranspiration. Principles and applications for water management. Oakville: Apple Academic Press. Güntner, A., Stuck, J., Werth, S., Döll, P., Verzano, K., Merz, B., 2007. A global analysis of temporal and spatial variations in continental water storage. Water Resour. Res. 43, W05416. Doi: 10.1029/2006WR005247. Häckel, H., 1999. Meteorologie. (4 ed.). Stuttgart: Ulmer. Harbeck, E., 1962. A practical technique for measuring reservoir evaporation utilizing mass- transfer theory. United States Geological Survey Professional Paper 272-E. Harding, R.J., Weedon, G.P., van Lanen, H.A.J., Clark, D.B., 2014. The future for global water assessment. J. Hydrol. 518, 186-193. Doi: 10.1016/j.jhydrol.2014.05.014. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from temperature. Appl. Eng. Agric. 1, 96-99. Doi: 10.13031/2013.26773. Hartman, H.L., Mutmansky, J.M., 2002. Introductory mining engineering. (2 ed.). Hoboken: John Wiley & Sons.

60

State of the art

Haude, W., 1955. Zur Bestimmung der Verdunstung auf möglichst einfache Weise. Dt. Wetterdienst, 11, 1-24. Haude, W., 1959. Die Verteilung der potentiellen Verdunstung in Ägypten. Erdkunde 13, 214-224. Doi: 10.3112/erdkunde.1959.03.04. Hauser, V.L., 2009. Evapotranspiration covers for landfills and waste sites. Boca Raton: CRC Press. Hauser, V.L., Gimon, D.M., Bonta, J.V., Howell, T.A., Malone, R.W., Williams, J.R., 2005. Models for hydrologic design of evapotranspiration landfill covers. Environ. Sci. Technol. 39, 7226- 7233. Doi: 10.1021/es048020e. Hauser, V.L., Weand, B.L., Gill, M.D., 2001. Natural covers for landfills and buried waste. J. Environ. Eng. 127, 768-775. Doi: 10.1061/(ASCE)0733-9372(2001)127:9(768). He, L., Matthew, C., Jones, C.S., Hatier, J.H.B., 2017. Productivity in simulated drought and post- drought recovery of eight ryegrass cultivars and a tall fescue cultivar with and without Epichloë endophyte. Crop Pasture Sci. 68, 176-187. Doi: 10.1071/CP16208. Heinrichs, H.U., Schumann, D., Vögele, S., Biß, K.H., Shamon, H., Markewitz, P., et al., 2017. Integrated assessment of a phase-out of coal-fired power plants in Germany. Energy 126, 285- 305. Doi: 10.1016/j.energy.2017.03.017. Henderson, T., 2017. The anatomy of a wave. http://www.physicsclassroom.com/class/waves/Lesson-2/The-Anatomy-of-a-Wave (accessed 18 November 2017). Hendrickson, J.R., Schmer, M.R., Sanderson, M.A., 2013. Water use efficiency by switchgrass compared to a native grass or a native grass alfalfa mixture. Bioenerg. Res. 6, 746-754. Doi: 10.1007/s12155-012-9290-3. Hendriks, M.R., 2010. Introduction to physical hydrology. Oxford: Oxford University Press. Hermsmeyer, D., 2001. Soil physical and hydrological evaluation of aluminum recycling by- product as an infiltration barrier for potash mine tailings (Doctoral Dissertation). Hanover University, Welfengarten. Hillel, D., 1998. Environmental soil physics: Fundamentals, applications, and environmental considerations. Cambridge: Academic Press. Hodnett, M.G., Tomasella, J., 2002. Marked differences between van Genuchten soil water- retention parameters for temperate and tropical soils. A new water-retention pedo-transfer functions developed for tropical soils. Geoderma 108, 155-180. Doi: 10.1016/S0016- 7061(02)00105-2. Hoffmann, M., Schwartengräber, R., Wessolek, G., Peters, A., 2016. Comparison of simple rain gauge measurements with precision lysimeter data. Atmos. Res. 174-175, 120-123. Doi: 10.1016/j.atmosres.2016.01.016.

61

State of the art

Holm, O., Simon, F., 2017. Innovative treatment trains of bottom ash (BA) from municipal solid waste incineration (MSWI) in Germany. Waste Manage. 59, 229-236. Doi: 10.1016/j.wasman.2016.09.004 Hopmans, J.W., 2010. Infiltration and unsaturated zone. In Wilderer, P.A. (Ed.). Treatise Water Sci. 103-114. Burlington: Elsevier Science Hopmans, J.W., Schoups, G., 2006. Soil water flow at different spatial scales. Encyclopedia of Hydrological Sciences. John Wiley & Sons, Ltd. Doi: 10.1002/0470848944.hsa070. Horton, R., Horn, R., Bachmann, J., Peth, S., 2016. Essential soil physics. An introduction to soil processes, functions, structure and mechanics. Stuttgart: Schweizerbart Science Publishers. Howard, J., 2017. Anthropogenic soils. Progress in Soil Science. Cham: Springer International Publishing. Inkaew, K., Saffarzadeh, A., Shimaoka, T., 2016. Modeling the formation of the quench product in municipal solid waste incineration (MSWI) bottom ash. Waste manage. 52, 159-168. Doi: 10.1016/j.wasman.2016.03.019. Jensen, K.B., Asay, K.H., Waldron, B.L., 2001. Dry matter production of orchardgrass and perennial ryegrass at five irrigation levels. Crop Sci. 41, 479-487. Doi: 10.2135/cropsci2001.412479x. Jensen, M.E., Burman, R.D., Allen, R.G., 1990. Evapotranspiration and irrigation water requirements. A manual. New York: American Society of Civil Engineers. Jones, J.W., Antle, J.M., Basso, B., Boote, K.J., Conant, R.T., Foster, I., et al., 2017. Brief history of agricultural systems modeling. Agric. Syst. 155, 240-254. Doi: 10.1016/j.agsy.2016.05.014. Kidd, C., Huffman, G., 2011. Global precipitation measurement. Meteorol. Appl. 18, 334-353. Doi: 10.1002/met.284. Kirkham, M.B., 2014. Principles of soil and plant water relations. (2 ed.). Amsterdam: Elsevier Academic Press. Kørup, K., Laerke, P.E., Baadsgaard, H., Andersen, M.N., Kristensen, K., Münnich, C., et al., 2017. Biomass production and water use efficiency in perennial grasses during and after drought stress. GCB Bioenergy 97. Doi: 10.1111/gcbb.12464. Lal, R., Shukla, M., 2004. Principles of soil physics. New York: Taylor and Francis. Lamb, E., 2015. A complete study of evapotranspiration. New York: Callisto Reference. Larcher, W., 2003. Physiological plant ecology. Ecophysiology and stress physiology of functional groups. (4 ed.). Berlin: Springer. Lascano, R.J., 1991. Review of models for predicting soil water balance. Soil water balance in the Sudano–Sahelian zone. Proceedings of the Niamey Workshop. IAHS Publ. No. 199. 443-458

62

State of the art

Leitinger, G., Ruggenthaler, R., Hammerle, A., Lavorel, S., Schirpke, U., Clement, J., et al., 2015. Impact of droughts on water provision in managed alpine grasslands in two climatically different regions of the Alps. Ecohydrol. 8, 1600-1613. Doi: 10.1002/eco.1607. Lim, W.H., Roderick, M.L., Hobbins, M.T., Wong, S.C., Farquhar, G.D., 2013. The energy balance of a US Class A evaporation pan. Agric. For. Meteorol. 182-183, 314-331. Doi: 10.1016/j.agrformet.2013.07.001. Linacre, E., 1994. Estimating U.S. Class A Pan evaporation from few climate data. Water Int. 19, 5-14. Doi: 10.1080/02508069408686189. Liu, X., Xu, C., Zhong, X., Li, Y., Yuan, X., Cao, J., 2017. Comparison of 16 models for reference crop evapotranspiration against weighing lysimeter measurement. Agric. Water Manage. 184, 145-155. Doi: 10.1016/j.agwat.2017.01.017. Loos, C., Gayler, S., Eckart, P., 2007. Assessment of water balance simulations for large-scale weighing lysimeters. J. Hydrol. 335, 259-270. Doi: 10.1016/j.jhydrol.2006.11.017. Lu, J., Sun, G., McNulty, S.G., Amatya, D.M., 2005. A comparison of six potential evapotranspiration methods for regional use in the Southeastern United States. JAWRA J. Am. Water Resour. Assoc. 41, 621-633. Doi: 10.1111/j.1752-1688.2005.tb03759.x. Marshall, S.J., 2014. The water cycle. In Reference module in earth systems and environmental sciences. Elsevier. Doi: 10.1016/B978-0-12-409548-9.09091-6. Meissner, R., Seeger, J., Rupp, H., Seyfarth, M., Borg, H., 2007. Measurement of dew, fog, and rime with a high-precision gravitation lysimeter. J. Plant Nutr. Soil Sci. 170, 335-344. Doi: 10.1002/jpln.200625002. Melo, G.L., Fernandes, A.L.T., 2012. Evaluation of empirical methods to estimate reference evapotranspiration in Uberaba, State of Minas Gerais, Brazil. Eng. Agríc. 32, 875-888. Doi: 10.1590/S0100-69162012000500007. Merten, G.H., Araújo, A.G., Biscaia, R.C.M., Barbosa, G.M.C., Conte, O., 2015. No-till surface runoff and soil losses in southern Brazil. Soil Tillage Res. 152, 85-93. Doi: 10.1016/j.still.2015.03.014. Meuser, H., 2013. Contaminated urban soils. Heidelberg: Springer. Mishra, S.K., Chaudhary, A., Shrestha, R.K., Pandey, A., Lal, M., 2014. Experimental verification of the effect of slope and land use on SCS Runoff Curve Number. Water Resour. Manage. 28, 3407-3416. Doi: 10.1007/s11269-014-0582-6. Moeletsi, M.E., Walker, S., Hamandawana, H., 2013. Comparison of the Hargreaves and Samani equation and the Thornthwaite equation for estimating dekadal evapotranspiration in the Free State Province, South Africa. Phys. Chem. Earth, Parts A/B/C 66, 4-15. Doi: 10.1016/j.pce.2013.08.003.

63

State of the art

Monteith, J.L., Unsworth, M.H., 1990. Principles of environmental physics. (2 ed.). London: Edward Arnold. Morbidelli, R., Saltalippi, C., Flammini, A., Cifrodelli, M., Corradini, C., Govindaraju, R.S., 2015. Infiltration on sloping surfaces. Laboratory experimental evidence and implications for infiltration modeling. J. Hydrol. 523, 79-85. Doi: 10.1016/j.jhydrol.2015.01.041. Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of unsaturated porous media. Water Resour. Res. 12, 513-522. Doi: 10.1029/WR012i003p00513. Narasimhan, T.N., 2009. Hydrological cycle and water budgets. In Likens, G.E., (Ed.). Encyclopedia of inland waters, 714-720. Amsterdam: Science Direct. Doi: 10.1016/B978- 012370626-3.00010-7. Nelles, M., Grünes, J., Morscheck, G., 2016. Waste management in Germany - Development to a sustainable circular economy? Procedia Environ. Sci. 35, 6-14. Doi: 10.1016/j.proenv.2016.07.001 Niessing, S., 2005. Begrünungsmaßnahmen auf der Rückstandshalde des Kaliwerkes - Sigmundshall in Bokeloh (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 25/2005, Universität Kassel, Witzenhausen. Nyhan, J.W., 2005. A seven-year water balance study of an evapotranspiration landfill cover varying in slope for semiarid regions. Vadose Zone J. 4, 466-480. Doi: 10.2136/vzj2003.0159. Oberholzer, S., Prasuhn, V., Hund, A., 2017. Crop water use under Swiss pedoclimatic conditions - Evaluation of lysimeter data covering a seven-year period. Field Crops Res. 211, 48-65. Doi: 10.1016/j.fcr.2017.06.003. Papke, G., Schmeisky, H., 2013. Rekultivierung von Rückstandshalden der Kaliindustrie. Ergebnisse aus langjährigen wissenschaftlichen Begleituntersuchungen der Begrünungsflächen auf der Kalirückstandshalde Sigmundshall in Bokeloh. Ökologie und Umweltsicherung, Bd. 35/2013, Universität Kassel, Witzenhausen. Paredes, M., 2016. The glocalization of mining conflict: Cases from Peru. Extr. Ind. Soc. 3, 1046- 1057. Doi: 10.1016/j.exis.2016.08.007 Park, J.H., Edraki, M., Mulligan, D., Jang, H.S., 2014. The application of coal combustion by- products in mine site rehabilitation. J. Cleaner Prod. 84, 761-772. Doi: 10.1016/j.jclepro.2014.01.049. Patil, N.G., Singh, S.K., 2016. Pedotransfer functions for estimating soil hydraulic properties. A review. Pedosphere 26, 417-430. Doi: 10.1016/S1002-0160(15)60054-6. Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences 193, 120-145.

64

State of the art

Pereira, A.R., Pruitt, W.O., 2004. Adaptation of the Thornthwaite scheme for estimating daily reference evapotranspiration. Agric. Water Manage. 66, 251-257. Doi: 10.1016/j.agwat.2003.11.003. Pereira, L.S.; Alves, I., 2013. Crop water requirements. Earth Systems and Environmental Sciences. Elsevier. Perez, A.A., 2016. The mineral industry of Germany. In US Geology Survey. 2013 Minerals Yearbook https://minerals.usgs.gov/minerals/pubs/country/2013/myb3-2013-gm.pdf (accessed 22 November 2017). Priestley, C.H.B., Taylor, R.J., 1972. On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Wea. Rev. 100, 81-92. Doi: 10.1175/1520- 0493(1972)100<0081:OTAOSH>2.3.CO;2. Radcliffe, D.E., Simunek, J., 2010. Soil physics with HYDRUS. Modeling and applications. Florida: CRC Press. Rast, M., Johannessen, J., Mauser, W., 2014. Review of understanding of earth’s hydrological cycle. Observations, theory and modelling. Surv. Geophys. 35, 491-513. Doi: 10.1007/s10712- 014-9279-x. Rauche, H., 2015. Die Kaliindustrie im 21. Jahrhundert. Stand der Technik bei der Rohstoffgewinnung und der Rohstoffaufbereitung sowie bei der Entsorgung der dabei anfallenden Rückstände. (1. Aufl.). Berlin: Springer. Rock, S., Myers, B., Fiedler, L., 2012. Evapotranspiration (ET) covers. Int. J. Phytorem. 14, 1-25. Doi: 10.1080/15226514.2011.609195. Rose, C.W., 2004. An introduction to the environmental physics of soil, water and watersheds. Cambridge: Cambridge University Press. Sabbas, T., Polettini, A., Pomi, R., Astrup, T., Hjelmar, O., Mostbauer, P., et al., 2003. Management of municipal solid waste incineration residues. Waste Manage. 23, 61-88. Doi: 10.1016/S0956- 053X(02)00161-7. Satheeshkumar, S., Venkateswaran, S., Kannan, R., 2017. Rainfall-runoff estimation using SCS- CN and GIS approach in the Pappiredipatti watershed of the Vaniyar sub basin, South India. Model. Earth Syst. Environ. 3, 24. Doi: 10.1007/s40808-017-0301-4. Saxton, K.E., Rawls, W.J., 2006. Soil water characteristic estimates by texture and organic matter for hydrologic solutions. Soil Sci. Soc. Am. J. 70, 1569-1578. Doi: 10.2136/sssaj2005.0117. Schaap, M.G., Leij, F.J., 2000. Improved prediction of unsaturated hydraulic conductivity with the Mualem-van Genuchten model. Soil Sci. Soc. Am. J. 64, 843-851. Doi: 10.2136/sssaj2000.643843x.

65

State of the art

Schaap, M.G., Leij, F.J., van Genuchten, M.Th., 2001. Rosetta: A computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. J. Hydrol. 251, 163-176. Doi: 10.1016/S0022-1694(01)00466-8. Scheer, T., 2001. Rekultivierung von Rueckstandshalden der Kaliindustrie. Untersuchungen zur Nutzbarkeit aufbereiteter Salzschlacke der Sekundaeraluminium-Industrie als Rekultivierungsmaterial einer Kali-Rückstandshalde (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 20/2001, Universität Kassel, Witzenhausen. Schmeisky, H., Hofmann, H., 2000. Rekultivierung von Rückstandshalden der Kaliindustrie - Untersuchungen zum Salzaustrag, zur Sukzession sowie Maßnahmen und Erkenntnisse zur Begrünung. Ökologie und Umweltsicherung, 19/2000, Universität Kassel, Witzenhausen Schnabel, W.E., Munk, J., Lee, W.J., Barnes, D.L., 2012. Four-year performance evaluation of a pilot-scale evapotranspiration landfill cover in Southcentral Alaska. Cold Reg. Sci. Technol. 82, 1-7. Doi: 10.1016/j.coldregions.2012.03.009. Seiler, K.P., Gat, J.R., 2007. Groundwater recharge from run-off, infiltration and percolation. Water Science and Technology Library, v. 55. Dordrecht: Springer. Sentelhas, P.C., Gillespie, T.J., Santos, E.A., 2010. Evaluation of FAO Penman–Monteith and alternative methods for estimating reference evapotranspiration with missing data in Southern Ontario, Canada. Agric. Water Manage. 97, 635-644. Doi: 10.1016/j.agwat.2009.12.001. Shukla, M., 2014. Soil physics. An introduction. Boca Raton: CRC Press. Shuttleworth, W.J., 2012. Terrestrial hydrometeorology. Hoboken: Wiley-Blackwell. Siebert, S., Burke, J., Faures, J.M., Frenken, K., Hoogeveen, J., Döll, P., Portmann, F.T., 2010. Groundwater use for irrigation - a global inventory. Hydrol. Earth Syst. Sci. 14, 1863-1880. Doi: 10.5194/hess-14-1863-2010. Simunek, J., Šejna, M., Saito, H., Sakai, M., van Genuchten, M.Th., 2013. The HYDRUS-1D software package for simulating the movement of water, heat, and multiple solutes in variably saturated media, Version 4.17. Department of Environmental Sciences, University of California Riverside, Riverside, California, USA. Simunek, J., van Genuchten, M.Th., Šejna, M., 2008. Development and applications of the HYDRUS and STANMOD software packages and related codes. Vadose Zone J. 7, 587-600. Doi: 10.2136/vzj2007.0077. Smith, M., 1992. CROPWAT. A computer program for irrigation planning and management. FAO Irrigation and Drainage Paper 46, Rome. Smith, M.W., Tice, A.R., 1988. Measurement of the unfrozen water content of soil: comparison of NMR and TDR methods. U.S. Army Cold Regions Research and Engineering Laboratory.

66

State of the art

Stancalie, G., Marica, A., Toulios, L., 2010. Using earth observation data and CROPWAT model to estimate the actual crop evapotranspiration. Phys. Chem. Earth, Parts A/B/C 35, 25-30. Doi: 10.1016/j.pce.2010.03.013. Stürmer, A., 2012. Umweltverschmutzung durch K+S: Die dunkle Seite des Börsenstars. Spiegel Online. Hamburg; Germany. http://www.spiegel.de/wirtschaft/unternehmen/salzabwaesser- von-k-s-deutschland-verstoesst-gegen-eu-richtlinie-a-845290.html (accessed 22 November 2017). Taiz, L., Zeiger, E., Moller, I.M., Murphy, A., 2015. Plant physiology and development. (6 ed.). Sunderland: Sinauer. Thornthwaite, C.W., 1948. An approach toward a rational classification of climate. Geogr. Rev. 38, 55-94. Doi: 10.2307/210739. Toldi, O.,Tuba, Z., Scott, P., 2009. Vegetative desiccation tolerance. Is it a goldmine for bioengineering crops? Plant Sci. 176, 187-199. Doi: 10.1016/j.plantsci.2008.10.002. Topp, G.C., Ferré, P.A., 2002. Water content. In Dane, J.H., Topp, G.C. (Eds.). Methods of soil analysis. Part 4. 417-545. SSSA Book Series No. 5. Madison: Soil Science Society of America. Topp, G.C., 2003. State of the art of measuring soil water content. Hydrol. Process. 17, 2993-2996. Doi: 10.1002/hyp.5148. Tucci, C.E.M., 2005. Modelos hidrológicos. (2 ed.). Associação Brasileira de Recursos Hídricos / ABRH. Porto Alegre: Editora da UFRGS, Tukimat, N.N.A., Harun, S., Shahid, S., 2012. Comparison of different methods in estimating potential evapotranspiration at Muda Irrigation Scheme of Malaysia. J. Agric. Rural Dev. Trop. Subtrop. 113, 77-85. Turc, L., 1954. Le bilan d'eau des sols - Relations entre les précipitations, l'évaporation et l'écoulement. Ann. Agron., 491-595, Versailles. U. S. Department of Agriculture (USDA), 1986. Urban hydrology for small Watersheds. Technical Release 55, United States Department of Agriculture. (2 ed.). https://www.nrcs.usda.gov/Internet/FSE_DOCUMENTS/stelprdb1044171.pdf (accessed 12 December 2017). van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892-898. van Genuchten, M.Th., 1992. On estimating the hydraulic properties of unsaturated soils. In van Genuchten, M.Th., Leij, F.J., Lund, L.J., (Eds.). Indirect methods for estimating the hydraulic properties of unsaturated soils. 1-14. Proceedings of the International Workshop, Riverside, CA. van Genuchten, M.Th., Leij, F.J., Yates, S.R., 1991. The RETC code for quantifying the hydraulic functions of unsaturated soils. Environmental Protection Agency Report, 600/2-91/065, U.S. Salinity Laboratory, 93. 67

State of the art

Vries, F.T., Brown, C., Stevens, C.J., 2016. Grassland species root response to drought. Consequences for soil carbon and nitrogen availability. Plant Soil 409, 297-312. Doi: 10.1007/s11104-016-2964-4. Wallace, J.M.; Hobbs, P.V., 2006. Atmospheric science. An introductory survey. (2 ed.). Amsterdam: Elsevier. Wang, H., Qin, J., Hu, Y., 2017. Are green roofs a source or sink of runoff pollutants? Ecol. Eng. 107, 65-70. Doi: 10.1016/j.ecoleng.2017.06.035. Warrick, A.W., 2002. Soil physics companion. Boca Raton: CRC Press. Warren, J.K., 2016. Evaporites. A geological compendium. (2 ed.). Heidelberg: Springer. Watanabe, K., Wake, T., 2009. Measurement of unfrozen water content and relative permittivity of frozen unsaturated soil using NMR and TDR. Cold Reg. Sci. Technol. 59, 34-41. Doi: 10.1016/j.coldregions.2009.05.011. Wedig, M., 2014. German mining industry overview. Mining Report 150, 90-93. Doi: 10.1002/mire.201400008. Wessman, H., Salmi, O., Kohl, J., Kinnunen, P., Saarivuori, E., Mroueh, U.M., 2014. Water and society. Mutual challenges for eco-efficient and socially acceptable mining in Finland. J. Cleaner Prod. 84, 289-298. Doi: 10.1016/j.jclepro.2014.04.026. Willmott, C.J., Rowe, C.M., Mintz, Y., 1985. Climatology of the terrestrial seasonal water cycle. J. Climatol. 5, 589-606. Doi: 10.1002/joc.3370050602. Winter, S., 2016. Gewisses Restrisiko. Der Spiegel 2016, 16.06.16., 40-43. http://www.spiegel.de/spiegel/print/d-144314347.html (accessed 02 September 2016). Xu, C.Y., Singh, V.P., 2000. Evaluation and generalization of radiation‐based methods for calculating evaporation. Hydrol. Process. 14, 339-349. Doi: 10.1002/(SICI)1099- 1085(20000215)14:2<339::AID-HYP928>3.3.CO;2-F. Xu, C.Y., Singh, V.P., 2002. Cross comparison of empirical equations for calculating potential evapotranspiration with data from Switzerland. Water Resour. Manage. 16, 197-219. Doi: 10.1023/A:1020282515975. Yeh, H., 2017. Comparison of evapotranspiration methods under limited data. 01. In Bucur, D. (Ed.) Current perspective to predict actual evapotranspiration. InTech, http://dx.doi.org/10.5772/intechopen.68495 (accessed 19 November 2017). Zhang, J., Lei, T., Qu, L., Chen, P., Gao, X., Chen, C., et al., 2017. Method to measure soil matrix infiltration in forest soil. J. Hydrol. 552, 241-248. Doi: 10.1016/j.jhydrol.2017.06.032. Zhang, L., Wang, J., Bai, Z., Lv, C., 2015. Effects of vegetation on runoff and soil erosion on reclaimed land in an opencast coal-mine dump in a loess area. Catena 128, 44-53. Doi: 10.1016/j.catena.2015.01.016.

68

State of the art

Zhang, W., Sun, C., 2014. Parametric analyses of evapotranspiration landfill covers in humid regions. J. Rock Mech. Geotech. Eng. 6, 356-365. Doi: 10.1016/j.jrmge.2013.12.005.

69

Chapter 3

3 Water Balance Assessment of Different Substrates on Potash Tailings Piles using Non- Weighable Lysimeters

Carolina Bilibioa*, Christian Schellerta, Stefanie Retza, Oliver Hensela, Helge Schmeiskyb, Daniel Uteauc, Stephan Pethc a Department of Agricultural and Biosystems Engineering - University of Kassel, Nordbahnhofstraße 1a, D-37213 Witzenhausen, Germany b Schmeisky Environmental Consultancy, Steinstrasse 21, D-37213 Witzenhausen, Germany c Department of Soil Science - University of Kassel, Nordbahnhofstraße 1a, D-37213 Witzenhausen, Germany *Corresponding author ([email protected])

3.1 Graphical abstract

3.2 Highlights  The water balance of eight lysimeters on a potash tailings pile were evaluated.  Four substrates were used to create a feasible environment for crop development.  Perennial grasses evapotranspirated up to 70.2 % of annual precipitation within the first years.  The average drainage was 271.2 mm in 2014 and 192.1 mm in 2015.  High evapotranspiration decreases brine drainage from potash tailing piles.

70

Chapter 3

3.3 Abstract Water balance is an important tool to evaluate water deficit or excess in crop systems. However, few studies have evaluated the water balance of vegetation grown on potash mining residues because the mining waste’s high sodium chloride levels hinder agricultural development. Therefore, this study aims to measure the water balance components in eight non-weighing lysimeters installed on a potash tailings pile in Heringen (Werra), Germany. These lysimeters were filled with different mixtures of household waste incineration slags and coal combustion residues, resulting in 4 different substrates with two repetitions. Manual seeding was performed using 65 % of perennial ryegrass (Lolium perenne L.), 25 % red fescue (Festuca rubra L.) and 10 % Kentucky bluegrass (Poa pratensis L.). Environmental conditions were monitored using an automatic weather station; ground-level and 1-m-high rain gauges. Precipitation and drainage were recorded weekly following the initial saturation of the lysimeters. Water balance components were determined for two hydrological years based on the expression: ET (mm) = P - D, where ET is evapotranspiration, P is precipitation and D is drainage. In addition, evapotranspiration was studied using the standard FAO Penman-Monteith equation and Haude's method. The lysimeter water balance measured in 2014 revealed an actual evapotranspiration rate of 66.4 % for substrate 1, 66.9 % for substrate 2, 65.1 % for substrate 3 and 64.1 % for substrate 4. In 2015, evapotranspiration ranged from 65.7 % for substrate 4 to 70.2 % for substrate 1. The FAO Penman-Monteith and Haude's evapotranspiration models are generally higher than the actual water use of the green coverage by 67 % and 23 %, respectively. The present study suggests that an evapotranspiration cover for potash tailings piles may decrease brine drainage from these piles and reduce soil and water contamination.

Keywords Evapotranspiration; Perennial grass; Drainage; Potash mining; Rain gauges; Crop coefficient

71

Chapter 3

3.4 Introduction Water balance is an important tool for evaluating water deficit or excess in crop systems (Soldevilla-Martinez et al., 2014; Fisher, 2012; Groh et al., 2015). The methods used to measure water balance are based on the principle of mass conservation, accounting for inputs, outputs and storage changes of a given element in the environment (Rose, 2004). The inputs used to measure water balance are irrigation, precipitation and capillary rise from groundwater. Evaporation, transpiration, drainage and surface runoff are the outputs (Blume et al., 2010a, 2010b; Meissner et al., 2010). Studies of soil water balance are facilitated by the use of lysimeters (Hillel, 1998; Beeson, 2011; Meissner et al., 2008; Hagenau et al., 2015; Klammler and Fank, 2014). Lysimeters are large containers filled with soil and placed in the field (Lal and Shukla, 2004). They can be classified as either weighing or non-weighing (Ehlers and Goss, 2003). Non-weighing lysimeters evaluate water volume balance, whereas weighing lysimeters estimate the water mass balance (Allen et al., 1991; Allen et al., 2011). Lysimeters must be deep enough to promote root development (Hagenau et al., 2015; Meissner et al., 2010), with a minimum area of 2 m2 to account for soil variability (Aboukhaled et al., 1982). Moreover, crop population, height, irrigation and fertilization should be similar between lysimeters and the surrounding area (Gebler et al., 2015). In addition, the cultivated field around the lysimeter should be large enough to avoid the oasis effect (Kirkham, 2014; Gebler et al., 2015; Potchter et al., 2008). Many studies have been conducted to evaluate water balance components using lysimeters. Lopez-Urrea et al. (2012) studied the evapotranspiration rate of Vitis vinifera L. in Spain using weighable lysimeters, Gonzalez-Talice et al. (2012) examined the monthly water demand of apple cultivars using drainage lysimeters in Chile, and Piouceau et al. (2014) investigated the evapotranspiration rate and crop coefficients of five species of three-year-old bamboo on Reunion Island using percolation-type lysimeters. In contrast, few studies have assessed crop water balance on potash tailings piles due to the high salt content of the mining residues (Schmeisky and Podlacha, 2000; Hermsmeyer, 2001; Hermsmeyer et al., 2002). Potash refers to several minerals that contain potassium, such as sylvite - the source of potassium chloride - and carnallite, which consists of potassium magnesium chloride (Manning, 2015). The main potash deposits are located in the Northern Hemisphere. Canada, Russia, Belarus and Germany have more than 80 % of the world's potash reserves (Ciceri et al., 2015). Additional reserves are found in Israel, China, Chile, United Kingdom, United States, Spain, Jordan and Brazil (IPI, 2016; Rawashdeh and Maxwell, 2014).

72

Chapter 3

Most potash deposits originate from seawater evaporation in large inland basins, which occurred millions of years ago (K+S KALI GmbH, 2016a). As the ratio of sodium to potassium in the oceans is 27:1 (Searls, 1992), the mining and processing of potash result in millions of tons of liquid and solid residues with high rates of sodium chloride (Papke and Schmeisky, 2013; Hart, 1989). The brines are injected into deep geological formations, oceans or water courses (Rauche, 2015). In contrast, the solid tailings, which contain up to 90 % sodium chloride, are backfilled in mining voids or heaped near potash facilities (European Comission, 2009). The main concern regarding the tailings piles in the short and long term is the leaching of sodium chloride by precipitation erosion. This precipitation erosion generates brines, which in Germany are collected in impermeable dykes near the dumps (European Commission, 2009). The brines are then stored in sealed reservoirs and disposed of in surface water courses or deep wells (European Commission, 2009). In Saskatchewan, Canada, the brines are evaporated in large storage ponds, and the excess is injected into natural saline formations (Reid and Getzlaf, 2004). Whereas in Sergipe, Brazil, the brines and tailings are disposed of in the South Atlantic Ocean (Rauche, 2015). Discharging brines from the tailings piles in Germany has raised several discussions with drinking water associations, local communities and the government, due to the constant increase of salt in rivers and groundwater (Winter, 2016). A prominent example is the contamination of the Werra River, located in central Germany, which has been salinized since 1901 due to the point and diffuse discharge of salt wastewater from potash mining (Braukmann and Böhme, 2011). Coring and Bäthe (2011) reported that the salt concentration in the Werra River reached 30 g/l of chloride between 1950 and 1990. However, as of late, the salt concentration in this river has decreased to 2.5 g/l, as a result of an official regulation imposed by the government in 1942 (Arle and Wagner, 2013). The salt contamination of the Werra River has reduced the richness level of aquatic vascular plants (Coring and Bäthe, 2011; Bäthe and Coring, 2011). In addition, large differences between the macro invertebrate assemblages of the upstream and downstream regions of the Werra River have been verified (Braukmann and Böhme, 2011). Brine disposal in the Werra River must decrease to comply with the European Water Framework Directive, which requires that all European rivers present good ecological and chemical status by 2027 (European Parliament, 2000; European Comission, 2012). In addition, by 2075, the Werra River must have freshwater characteristics (K+S KALI GmbH, 2016b). In light of this, the University of Kassel and the Schmeisky Environmental Consultancy have been trying to develop an evapotranspiration cover to install over potash tailings piles in order of the K+S KALI GmbH (Papke and Schmeisky, 2013). An evapotranspiration cover includes the use of a soil layer covered with native grasses (Hauser, 2009). The soil works as a water reservoir, and the evaporation and transpiration of the crops empty this reservoir, effectively decreasing water infiltration into the wastes (Hauser, 2009). In addition to these characteristics, an evapotranspiration cover for potash 73

Chapter 3 tailings piles must be sufficiently stable to endure the approximately 38-degree pile slopes (Niessing, 2005). Moreover, there must not be any wind or water erosion, the pile must show a high infiltration rate and water retention capacity, and there can be no hazard to the public (Schmeisky and Papke, 2012). Different soil substitutes have been tested as evapotranspiration covers for potash tailings piles, such as soil, ashes and even an aluminum-recycling by-product. The use of an aluminum-recycling by-product blended with ashes was studied by Hermsmeyer (2001) and Scheer (2001) in lysimeter studies conducted in Wunstorf, Germany. Although the authors found a decreasing trend for the drainage, there is a need for salt leaching to support plant growth (Hermsmeyer, 2001; Hermsmeyer et al., 2002). The use of soil combined with fly ashes was tested by Podlacha (1999) using 8 lysimeters in Bleicherode, Germany. The author measured an average evapotranspiration rate of 78-83 %. However, in another study, Papke and Schmeisky (2013) highlight that soil is unsuitable due to the slopes of the tailings piles. Moreover, huge amounts of soil would be needed to cover the piles, which is unfeasible (Niessing, 2005). Previous studies also included the use of static layers over the tailings piles, such as synthetic mats and concrete; nevertheless, a static layer was shown to be unsuitable due to the surface and subsurface instability of the piles (Schmeisky and Hofmann, 2000). Another possibility considered the construction of a 400-km subsurface pipeline to inject the brines directly into the North Sea, which proved to be impracticable and unaffordable; moreover, it was not passed by the government (Bebenburg, 2014). Using the results from these past experiments, the present study seeks to further the research into evapotranspiration covers for potash tailings piles. Specifically, this article focuses on the water balance components of 8 non-weighing lysimeters filled with different mixtures of household waste incineration slags and coal combustion residues. The lysimeter field was installed on the tailings pile from the Wintershall production facility, which belongs to K+S KALI GmbH. The use of household waste incineration slags and coal combustion residues was considered because they are affordable and highly available in the study region. Specifically, more than 6 million tons of household waste incineration slags are produced annually in Germany (Alwast and Riemann, 2010). These residues are used in construction, roads, exported to other European countries or landfilled (Alwast and Riemann, 2010). In addition, more than 100 million tons of residues from coal combustion are generated in the European Union (Spliethoff, 2010). These are used for concrete production, construction of underground mines and mine recultivation (Feuerborn, 2011). The experimental site included a weather station, 4 rain gauges installed on the ground level and 5 rain gauges installed at 1-m height to monitor meteorological conditions (Hensel et al., 2014; Bilibio et al., 2015). The weather station recorded the mean values of air temperature, air humidity, 74

Chapter 3 wind speed, solar radiation, precipitation and Haude's potential evapotranspiration (Haude, 1955). Using the data recorded by the weather station, the FAO Penman-Monteith reference crop evapotranspiration was estimated (Allen et al., 1998). The main findings of the present study are the drainage rate and the evapotranspiration potential of a revegetation cover composed of household waste incineration slags and coal combustion residues. Although the experimental site remains intact, this article reports the results from the first two hydrological years, i.e., 2014 and 2015. The outcomes of the present research can foster advancements in the rehabilitation of potash tailing piles across the world, especially where precipitation exceeds the evaporation capacity of the tailings piles.

3.5 Materials and methods

3.5.1 Description of the experimental area The experiment was carried out on the tailings pile from the Wintershall production facility, which is part of the Werra combined potash mine of K+S KALI GmbH (Figure 3-1). The pile is located at 50° 53' 59.644" North and 9° 59' 10.079" East, on the outskirts of Heringen, Germany, and is also known as “Monte Kali” (Bartsch and Fröhlingsdorf, 2009; Konopka, 2015). It is approximately 240 m high and is composed of more than 200 million tons of potash tailings, deposited over a period of 40 years. At present, 22,000 tons of mining residues are added daily (Werra-Kalibergbau-Museum, 2016).

Figure 3-1: Part of the potash tailings pile named “Monte Kali” from the Wintershall production facility which belongs to K+S KALI GmbH. The pile is in the outskirts of Heringen, Germany. The experimental site was situated in the proximities of the conveyor belts which transport the tailings from the processing facilities to the heap (Picture from K+S KALI GmbH) 75

Chapter 3

The climate in Heringen (Werra) is classified as Cfb (cold, no dry season, temperate summer and four months over 10 °C) under the Köppen-Geiger classification (Schwarz, 2016; Peel et al., 2007; Kottek et al., 2006). According to the climatological normals for the period 1961-1990, Heringen has an average annual temperature of 8.5 oC and an average annual precipitation of 684 mm (Deutsche Wetterdienst, 2016; Lamprecht, 2016).

3.5.2 Experimental design and installation of lysimeters Lysimeters and meteorological devices were installed in July 2013. The total experimental field area covered 544 m² (27.2 x 20 m), with eight lysimeters. The lysimeters were 3-m deep and covered an area of 2 m². The experiment was divided into 4 treatments with two repetitions (Schmeisky and Papke, 2013). Different proportions of industrial wastes were used to fill the lysimeters, as follows: Treatment 1: 80 % household waste incineration slags (0-12 mm diameter), 20 % coal combustion residues. Treatment 2: 70 % household waste incineration slags (0-12 mm diameter), 30 % coal combustion residues. Treatment 3: 60 % household waste incineration slags (0-12 mm diameter), 10 % washed sand from gravel extraction, 30 % coal combustion residues. Treatment 4: 50 % household waste incineration slags (0-12 mm diameter), 30 % coal combustion residues, 10 % furnace bottom ashes with particle sizes between 0.2 and 2 mm, labelled “Kesselsand”, 10% original bottom ashes with particle sizes from 0 to 6.3 mm, labelled “Feinasche”. Organic compost with a diameter of 0-20 mm was applied 0.3 m from the lysimeters’ surface area, as well as over the entire experimental field, totaling 200 tons per hectare (Schmeisky and Papke, 2013). In addition, different fractions of gravel were placed on the bottom of the lysimeters to prevent the substrates from washing out and to facilitate the drainage of percolated water (Figure 3-2). Thus, the lysimeters contained 0.3 m (0-0.3 m) of substrate plus compost, 2.3 m (0.3-2.6 m) of the substrate mixture, and 0.4 m (2.6-3.0 m) of a drainage layer, containing sand with a diameter of 1.0-1.6 mm, fine gravel with a diameter of 2.0-3.1 mm, medium gravel with a diameter of 3.1-5.6 mm and coarse gravel with a diameter of 5.6-8.0 mm (Schmeisky and Papke, 2013). As observed in Figure 3-2, the lysimeters were installed over a potash tailings layer at a height of approximately 1 m with a 3 % slope to facilitate the outflow of percolated water. After filling the lysimeters, they were saturated and remained covered for 50 days to reach field capacity (Schmeisky and Papke, 2013).

76

Chapter 3

Figure 3-2: An example of the non- weighable lysimeter with 3-m deep and covering an area of 2 m2. The surface of the lysimeter included the substrate mixture plus organic compost. The subsurface layer contained the different portions of household waste incineration slags and coal combustion residues. Whereas the filter layer incorporated different fractions of gravel in order to avoid the washing out of the substrates The relationship between water content (ϴ) and the matric potential (ψ) of substrate 1 and substrate 4 at 0.20-0.26 m and 0.40-0.46 m depth was assessed in 2014. The water retention curve was adjusted according to the van Genuchten (1980) model (Eq. 3-1) and the soil water retention parameters were obtained with RETention Curve software (RETC, van Genuchten et al., 1991), Table 3-1.

휃푠 − 휃푟 (3-1) 휃(휓푚) = 휃푟 + ( 푛 푚) [1 + ((훼휓푚) )]

In which θ(ψm) is the association between water content (cm³/cm³), and matric potential (cm),

θr is the residual water content (cm³/cm³), θs is the water content at soil saturation (cm³/cm³), ψm matric potential of the soil (cm), α, m and n = fitting parameters. 77

Chapter 3

Table 3-1: Water retention curve parameters according to van Genuchten (1980) model of the substrates 1 and 4 Substrate Depth θr θs α m n R2 m ------cm3/cm3 ------1/cm Substrate 0.20-0.26 0.000 0.41778 0.03722 0.20951 1.26504 0.9864 1 0.40-0.46 0.000 0.43810 0.04962 0.18951 1.23382 0.9939 Substrate 0.20-0.26 0.000 0.43408 0.03817 0.18440 1.22609 0.9818 4 0.40-0.46 0.000 0.41436 0.02500 0.20506 1.25795 0.9950

The moisture volume (%) released by gravity, as well as the available and unavailable plant water were estimated using field capacity: pF 1.8; permanent wilting point: pF 4.2; total pore volume: ≅ 휃푠, Table 3-2.

Table 3-2: Water volume (%) released by gravity, available and unavailable moisture in substrates 1 and 4 Substrate 1 Substrate 4 Depth (m) 0.20-0.26 0.40-0.46 0.20-0.26 0.40-0.46 % released 11.4 11.7 8.3 8.1 % available 20.3 20.3 19.1 23.1 % unavailable 8.0 10.9 13.7 9.2

The lowest water volume level released by gravity was observed in substrate 4, circa 8.1 %, whereas the maximum was 11.7 %, registered in substrate 1. The available plant water ranged from 19.1 %-23.1 %. In addition, the unavailable plant water varied between 8.0 % and 13.7 %. The dry bulk density ranged from 1.18 g/cm3; in substrate 1 to 1.22 g/cm3; in substrate 4. The analysis of the particle size distribution from the fine fraction (Ø < 2 mm) showed that substrates 1-4 are classified as sandy loam (Blume et al., 2016), with on average 52 % of sand-size particles (0.063-2 mm), 43 % of silt-size particles (2-63 µm) and 5 % of clay-size particles (<2 µm). In addition, a coarse fraction of 42 % (Ø > 2 mm) was found. Further studies are being conducted to determine the physical properties of the substrates and will be addressed in future publications.

3.5.3 Drainage Discharge lines connected to the lysimeters drained percolated water. These lines were linked to 60-L barrels placed in a shelter nearby. The amount of drained water was first recorded on 26 July 2013 and was then measured weekly, on Thursdays between 9 a.m. and 10 a.m.

78

Chapter 3

3.5.4 Seeding and fertilization A seed mixture containing 65 % perennial ryegrass (Lolium perenne L.), 25 % red fescue (Festuca rubra L.) and 10 % Kentucky bluegrass (Poa pratensis L.) was used from 5 August to 26 September 2013, totaling 70 g/m2 (Schmeisky and Papke, 2013). In addition, the yearly total amount of fertilizer was 83 g/m2 in 2013, 193 g/m2 in 2014 and 94 g/m2 in 2015, consisting of 40.6 g/m2 of nitrogen, 55.4 g/m2 of phosphorus, 57.6 g/m2 of potassium and 6.6 g/m2 of magnesium (Schmeisky and Papke, 2013).

3.5.5 Meteorological data Micrometeorological data were collected automatically at 10-min intervals by the Thies-Clima weather station located at the experimental site (Hensel et al., 2014). The weather station analysis was performed by assessing the precipitation (mm, 1-m height), wind speed (m/s, 3-m height), air temperature (oC, 2-m height), soil temperature (oC, 0.3-m depth), relative air humidity (%, 2-m height) and solar radiation (W/m2, 2-m height). The data were sent by radio to the K+S KALI GmbH headquarters located in Heringen (Werra) (Hensel et al., 2014; Bilibio et al., 2015).

3.5.5.1 Precipitation The weather station measured precipitation using an ombrometer with a collection area of 200 cm2 (Hensel et al., 2014). In addition, precipitation was measured using 4 ground-level gauges (Braunisch, 2008) and 5 rain gauges installed at 1-m height (Figure 3-3).

a b c

Figure 3-3: Ombrometer from Thies weather station (a), ground-level rain gauge (b) and 1-m-high rain gauge (c) Ground-level and 1-m-high rain gauges had a collection area of 100 cm2 and 500 ml of storage capacity. Additionally, a porcelain sieve diameter was used between the collection funnel and the storage recipient to minimize evaporation and the entry of dirt particles (Hensel et al., 2014). Rain gauges installed on the ground level were located in the center of a pit with a 50-cm diameter and 79

Chapter 3

40-cm height to avoid boundary effects, such as splashing water (Hensel et al., 2014). Precipitation in gauges located at 1-m height and on the ground level was assessed weekly, on Thursdays (Hensel et al., 2014; Bilibio et al., 2015).

3.5.6 Determination of water balance components The actual evapotranspiration was determined as residual term of the simplified water balance expression (Aboukhaled et al., 1982; Bilibio et al., 2011; Bethune et al., 2008):

퐸푇푎 = 푃 − 퐷 (3-2) Where ETa is the actual evapotranspiration (mm), P is the precipitation (mm), and D is the drainage (mm). Precipitation was measured using ground-level gauge records. This approach is recommended to evaluate the water balance as the gauges are located at the same height as the lysimeters and are exposed to the same environmental conditions (Dietrich et al., 2016; Gebler et al., 2015; Klammler and Fank, 2014). The weekly volumetric drainage collected from the lysimeters was considered as outgoing water flux. Moreover, the actual evapotranspiration estimated using Eq. 3-2 was also studied in terms of adjusted crop coefficient, which distinguishes field evapotranspiration from grass reference evapotranspiration through the expression (Allen et al., 1998; Fisher, 2012):

퐸푇푎 (3-3) 퐾푐푎푑푗 = 퐸푇0

Where Kcadj is the adjusted crop coefficient (dimensionless), ETa is the actual evapotranspiration

(mm), and ET0 is the reference crop evapotranspiration (mm). Since a lysimeter is a confined volume of soil within a container with protruding rims, runoff can be neglected when determining the water balance (Aboukhaled et al., 1982; Phogat et al., 2013). In addition, changes in soil water storage become negligible in periodically saturated lysimeters (Lal and Shukla, 2004; Kirkham, 2014).

3.5.7 Reference crop evapotranspiration Reference evapotranspiration refers to a hypothetical surface covered by grass with a height of 0.12 m, a fixed surface resistance of 70 s/m and an albedo of 0.23 (Allen et al., 1998). As reference grass has no water or nutrient limitations, the only determining factor of reference evapotranspiration is the weather, which varies according to location and time of year (Allen et al., 1998).

The monthly reference crop evapotranspiration was assessed using the ET0 calculator program (FAO, 2014), which is based on the FAO Penman-Monteith equation (Allen et al., 1998):

80

Chapter 3

900 ( ) ( ) (3-4) 0.408 ∆ Rn - G + γ T+273 u2 es - ea 퐸푇0 = ∆ + γ (1 + 0.34 u2)

Where ET0 is the reference evapotranspiration (mm/day), Rn is net radiation at the crop surface (MJ/m2/day), G is soil heat flux density (MJ/m2/day), T is mean daily air temperature at 2-m height

(°C), u2 is wind speed at 2-m height (m/s), es is saturation vapor pressure (kPa), ea is actual vapor pressure (kPa), es-ea is saturation vapor pressure deficit (kPa), Δ is the slope of the vapor pressure curve (kPa/°C), and γ is the psychrometric constant (kPa/°C). After estimating the reference evapotranspiration, the crop evapotranspiration under standard conditions (ETc) was calculated using the following expression (Allen et al., 1998):

퐸푇푐 = 퐸푇0 퐾푐 (3-5) The crop coefficients recommended for ryegrass hay, presented by Allen et al. (1998) were considered: initial stage (Jan-Mar): 0.95; mid-season stage (Apr-Oct): 1.05; and late-season stage (Nov-Dec): 1.0 (Groh et al., 2015).

Reference and crop evapotranspiration (ET0, ETc) were compared with the potential evapotranspiration based on Haude´s method (Haude, 1955). Haude´s potential evapotranspiration was provided by the Thies weather station and is written as (Häckel, 1999; Loos et al., 2007):

퐸푇푝퐻푎푢푑푒 = χ (푒푠 − 푒푎)14 ℎ (3-6)

Where χ is the Haude factor according to the specific crop and es-e is the vapor pressure deficit in hPa for the temperature at 2 p.m.

3.5.8 Evaluation period and statistical analysis An evaluation of the meteorological data and water balance components was conducted for two hydrological years, 2014 and 2015, and their respective winter and summer seasons. In Germany, according to DIN 4049, the hydrological year begins on November 1st and ends on October 31st of the next calendar year, whereas the winter season ranges from November to April and the summer season runs from May to October (Deutsches Institut für Normung e.V, 1996). The period of installation, saturation, vegetation cover establishment and stabilization of substrates from July to October 2013 were defined as the previous phase. Descriptive statistics were used to summarize the meteorological data and water balance components. The central tendency of the data was determined using mean values, whereas the variability of the mean was determined with standard deviation (SD) and coefficient of variation (CV) (Crawley, 2014; Field et al., 2012; Couto et al., 2013).

81

Chapter 3

3.6 Results and discussion

3.6.1 Weather condition The annual, biennial and monthly precipitation values recorded during the two hydrological years at the experimental site, collected using the different gauges, are shown in Table 3-3. The climate normal from 1961 to 1990 for precipitation is also presented. The total precipitation verified at the experimental field in 2014 by the Thies weather station (778.8 mm), using ground-level (788.3 mm) and 1-m-high gauges (710.0 mm), was on average 11 % higher than the climatological normal of 684 mm documented for Heringen from 1961 to 1990 (Lamprecht, 2016). In 2015, the Thies weather station recorded 508.1 mm of precipitation, the ground-level gauges registered 584.4 mm, and the 1-m-high gauges, 484.9 mm. These values were on average 23 % lower than the historical average precipitation (Lamprecht, 2016).

Table 3-3: Precipitation (mm) at the Heringen experimental site during two hydrological years and climatological normals

Precipitation Precipitation Precipitation 1-m- Climate normal weather station ground-level gauges high gauges for precipitation

(n = 1) (n = 4) (n = 5) mm mm mm mm Water year / Month 2014 2015 2014 2015 2014 2015 1961 - 1990 Nov 69.8 15.4 78.1 18.1 66.4 15.1 55.0 Dec 37.3 47.4 32.1 60.7 24.0 52.7 66.0 Jan 36.6 50.2 43.8 68.5 33.9 54.2 54.0 Feb 16.8 12.8 16.1 55.6 14.6 14.8 43.0 Mar 12.6 18.4 12.4 22.4 11.9 18.1 51.0 Apr 33.7 47.5 30.8 54.8 27.8 48.2 54.0 Mai 86.0 14.8 82.6 13.3 75.9 11.5 67.0 Jun 45.9 45.0 46.6 42.0 42.6 36.4 77.0 Jul 221.6 96.3 237.6 88.8 216.3 85.3 60.0 Aug 71.9 56.0 67.6 52.8 62.4 54.3 60.0 Sep 87.0 67.2 81.3 65.5 80.9 59.4 49.0 Oct 59.6 37.1 59.4 41.8 53.3 34.8 48.0 Winter (Ʃ) 206.8 191.7 213.3 280.1 178.6 203.1 323.0 Summer (Ʃ) 572.0 316.4 575.0 304.3 531.4 281.8 361.0 Water year (Ʃ) 778.8 508.1 788.3 584.4 710.0 484.9 684.0

The highest precipitation was registered in summer 2014, approximately 74 % of the annual precipitation, due to the precipitation observed in July 2014 (225.2 mm), which was 275 % higher

82

Chapter 3 than the historical average. The monthly analysis of precipitation also revealed that 13.2 mm was found in May 2015, which is 19.7 % of the historical average for the same period. The evaluation of the gauges showed that ground-level gauges registered 1.2 % more precipitation than the Thies weather station and 11.0 % more than the 1-m-high gauges in 2014. In 2015, the total rainfall recorded by the ground-level gauges was 15.0 % higher than the value documented by the Thies weather station and 20.5 % higher than 1-m-high gauges. Since wind speed increases with height (Geiger et al., 2009; Häckel, 1999), less wind interfering with the ground-level gauges may result in higher precipitation amounts, mainly during winter (World Meteorological Organization, 2008; Nolz et al., 2014; Gebler et al., 2015). The precipitation obtained from the ground-level gauges in winter 2014 (213.3 mm) was 3.1 % higher than the precipitation verified by the Thies weather station and 19.4 % higher than the 1-m-high gauges. This same phenomenon was observed in winter 2015, where ground-level gauges registered 46.1 % more precipitation than the Thies weather station and 37.9 % more than the 1-m-high gauges. However, it is important to highlight that ground-level gauges were sometimes covered with snow in winter (31.12.2014; 05 and 12.02.2015), which may have skewed precipitation readings (Groh et al., 2015). Lower differences were measured in summer 2014 and summer 2015, where ground- level gauges recorded higher average values of 4.4 % and 2.1 %, respectively, compared with the other gauges. The average air temperature observed in 2014 (10.6 oC) and in 2015 (9.2 oC) was higher than the historical mean for Heringen, i.e., 8.5 oC (Lamprecht, 2016). Further analyses revealed higher mean temperatures for summer, i.e., 14.8 oC in 2014 and 14.7 oC in 2015, compared with winter, i.e., 5.6 oC in 2014 and 3.7 oC in 2015. Hunt and Easton (1989) found that the optimum temperature for perennial ryegrass ranges from 18 to 20 oC (Blombäck and Eckersten, 1997; Monteith, 1977; Norris, 1985; Wherley and Sinclair, 2009). In addition, Allen et al. (1998) and Hillel (1998) argue that air temperature increases because of solar radiation absorption and heat emission from the earth, which in turn is transferred to crops and increases evapotranspiration. Monthly and biennial mean values of the mean air temperature are shown in Figure 3-4. A similar pattern was observed for soil temperature. The mean soil temperature recorded in 2014, i.e., 11.0 oC, was similar to that in 2015, i.e., 10.2 oC. The values were higher in the summer, i.e., 16.3 oC in 2014 and 15.6 oC in 2015, and lower in winter, i.e., 5.7 oC in 2014 and 4.9 oC in 2015 (Figure 3-4). Soil temperatures affect root and shoot growth as well as the physical, biological and chemical processes in soils (Hauser, 2009; Brady and Weil, 2014). Feldhake and Boyer (1986) studied the effects of soil temperature on evaporation using C3 cool season grasses, such as tall fescue (Festuca arundinacea Schreb.) and orchardgrass (Dactylis glomerata L.) and C4 warm season grasses, i.e., bermudagrass (Cynodon dactylon L.) and buffalograss (Buchloe dactyloides (Nutt.) Engelm.). The authors verified that evapotranspiration increased as soil temperature 83

Chapter 3 increased from 13 to 29 oC, except for orchardgrass at 29 oC. According to Lyons et al. (2007) and Brady and Weil (2014), the optimum soil temperature for most C3 grasses, such as perennial ryegrass, ranges from 10 to 18 oC.

30 30

25 25 C) C) o C) C) o 20 20

15 15

10 10 Mean ( Mean soil temperature

Mean ( Mean air temperature 5 5

0 0 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 Winter Summer Winter Summer Winter Summer Winter Summer 2014 2015 2014 2015

Time (water years) Time (water years)

100 250

80 200

60 150

40 100

50

20 radiation (W/m²) Mean solar Mean relative air humidity humidity (%) Mean relative air

0 0 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 Winter Summer Winter Summer Winter Summer Winter Summer 2014 2015 2014 2015

Time (water years) Time (water years)

84

Chapter 3

14 160

12 140 (mm) 0 120 ET

10 - 100 8 80 6 60 4 Mean wind speed (m/s)Mean wind 40

2 20

0 0 evapotranspiration Reference 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 Winter Summer Winter Summer Winter Summer Winter Summer 2014 2015 2014 2015

Time (water years) Biennial means Monthly sums

Figure 3-4: Monthly and biennial mean values of the mean air temperature, substrates temperature, relative air humidity, solar radiation, wind speed and reference evapotranspiration from 2014 to 2015 On the one hand, air humidity enables direct crop absorption of water vapor and increases leaf photosynthesis under high light intensities (Chang, 2009). On the other hand, lower evapotranspiration rates are observed in regions with high and constant air humidity as the air is close to saturation (Allen et al., 1998). In this study, the mean relative air humidity was 82.4 % in 2014 and 80.0 % in 2015. Moreover, the relative air humidity was inversely related to temperature, i.e., lower values in summer and higher values in winter (Figure 3-4). This phenomenon can be explained by the increasing capacity of the air to retain water vapor with rising temperatures, which consequently increase the saturated vapor pressure (Allen et al., 1998). Nevertheless, relative air humidity is inversely proportional to saturated vapor pressure (Allen et al., 1998). The weather analysis also included the solar radiation. Solar radiation is the main source of energy in the process of water vaporization (Hillel, 1998; Allen et al., 1998) and is responsible for dry matter production during photosynthesis (Chang, 2009; Monteith, 1977). The mean value was 118.9 W/m2 in 2014 and 127.0 W/m2 in 2015. Mean solar radiation was higher in summer, i.e., 168.4 W/m2 in 2014 and 180.6 W/m2 in 2015, and lower in winter, i.e., 69.4 W/m2 in 2014 and 73.4 W/m2 in 2015. Further weather analyses showed that the vapor pressure deficit (VPD), i.e., the difference between saturated and actual vapor pressure, or the atmospheric evaporative demand (Fletcher et al., 2007; Chang, 2009), was 0.37 kPa in 2014 and 0.41 kPa in 2015. As the saturated vapor pressure increases exponentially with temperature (Sinclair et al., 2007), VPD values were higher in the summer season, i.e., 0.52 kPa in 2014 and 0.65 kPa in 2015, and lower in winter, i.e., 0.21 kPa in 2014 and 0.18 kPa in 2015. A limit to maximum crop transpiration is observed at a VPD of ~2 kPa;

85

Chapter 3

above this threshold, transpiration and CO2 assimilation decrease due to a reduction in stomatal conductance (Wherley and Sinclair, 2009; Fletcher et al., 2007; Sinclair et al., 2007). Wind is responsible for removing water vapor from above the evaporative surface and replacing it with drier air (Allen et al., 1998), consequently affecting the transpiration rate and CO2 intake (Chang, 2009). At a height of 2-m, wind can be classified as light (≤ 1.0 m/s), light to moderate (2.0 m/s), moderate to strong (4.0 m/s), or strong (≥ 5.0 m/s) (Allen et al., 1998). By evaluating the wind speed recorded by the weather station, it was verified that wind speed in 2015, i.e., 3.1 m/s, was higher than in 2014, i.e., 2.8 m/s. With the location of the experiment and the weather measurements, such as air temperature, solar radiation, air humidity, and wind speed, the water demand of a hypothetical grass was estimated. The monthly average reference evapotranspiration (FAO, 2014) was 58.8 mm in 2014 and 64.7 mm in 2015. Higher evapotranspiration was calculated for summer, i.e., 85.9 mm/month in 2014 and 99.7 mm/month in 2015, while the lowest values were calculated for winter, i.e., 31.7 mm/month in 2014 and 29.6 mm/month in 2015. Total reference evapotranspiration was 705.3 mm in 2014, of which 190.0 mm (26.9 %) occurred in winter and 515.3 mm (73.1 %) in summer. In 2015, total reference evapotranspiration was 776.3 mm, of which 177.8 mm (22.9 %) was registered in winter and 598.5 mm (77.1 %) in summer.

Additional analysis of ryegrass evapotranspiration (ETc), estimated from reference evapotranspiration (ET0) and the standard crop coefficients (Kc), amounted to 730.4 mm in 2014 (mean of 60.9 mm/month) and 807.1 mm in 2015 (mean of 67.3 mm/month). It is important to note that ETc in 2014 was close to the precipitation level, i.e., 788.3 mm; however, ETc in 2015 was higher than the total precipitation recorded using the ground-level gauges, i.e., 584.4 mm. A closer examination of crop evapotranspiration in 2015 showed that crop water requirement in winter was 178.7 mm compared with 628.4 mm in summer. In contrast, total precipitation measured in winter 2015 was 280.1 mm compared with 304.3 mm in summer 2015. Hence, one can observe that summer precipitation in 2015 supplied only 48.4 % of the standard crop water use. Potential evapotranspiration recorded by the Thies weather station using Haude's method (Haude, 1955) showed similar means between 2014 (42.3 mm/month) and 2015 (47.6 mm/month). Summer had a higher potential evapotranspiration, i.e., 61.4 mm/month in 2014 and 75.5 mm/month in 2015, compared with the lowest averages in winter, i.e., 23.2 mm/month in 2014 and 19.8 mm/month in 2015. Evapotranspiration calculated using Haude's method was 28 % lower than the

ET0 and 30.5 % lower than ETc in 2014. Similarly, in 2015, evapotranspiration calculated using

Haude's method was 26.4 % and 29.2 % lower than ET0 and ETc, respectively.

86

Chapter 3

3.6.2 Water balance

3.6.2.1 Drainage The annual, biennial and monthly drainage of substrates 1, 2, 3 and 4 can be observed in Table 3-4. The mean drainage registered for substrate 1 was 22.1 mm/month in 2014 and 14.5 mm/month in 2015. The highest monthly drainages were recorded in winter 2014 (mean of 29.8 mm), followed by winter 2015 (mean of 24.8 mm) and summer 2014 (mean of 14.4 mm). The lowest drainage was measured in summer 2015 (4.2 mm/month). The total drainage of substrate 1 was 265.2 mm in 2014, i.e., approximately 33.6 % of the annual ground-level precipitation, and 173.9 mm in 2015 (29.8 % of annual rainfall). In 2014, 67.5 % of the total discharge occurred during the winter season, and in 2015, this value was 85.6 %. Substrates 2, 3 and 4 showed a similar trend.

Table 3-4: Drainage of lysimeters according to different substrates at the Heringen experimental site during the two hydrological years

Substrate 1 Substrate 2 Substrate 3 Substrate 4

(n = 2) (n = 2) (n = 2) (n = 2) mm mm mm mm Water year / Month 2014 2015 2014 2015 2014 2015 2014 2015 Nov 68.0 14.1 70.2 13.6 66.6 15.5 63.1 14.9 Dec 41.5 18.4 40.5 23.5 41.7 15.9 38.6 15.9 Jan 37.5 39.7 40.9 51.7 34.7 50.4 32.0 54.4 Feb 21.0 27.5 21.4 28.8 19.7 31.7 17.4 29.4 Mar 9.4 20.0 8.7 21.5 12.2 25.0 10.0 23.7 Apr 1.6 29.3 1.2 35.7 3.1 38.8 3.3 39.0 Mai 0.4 10.8 0.4 9.3 0.3 14.5 1.3 10.8 Jun 0.1 2.4 0.2 1.5 0.3 4.3 1.1 2.6 Jul 18.3 0.6 31.8 0.7 30.6 0.6 40.5 0.1 Aug 21.5 2.3 12.2 2.9 20.0 0.1 25.3 1.0 Sep 25.5 1.4 18.9 2.3 23.3 0.0 28.6 2.9 Oct 20.4 7.5 14.8 5.9 22.8 0.0 21.8 5.7 Winter (Ʃ) 179.0 148.9 183.0 174.8 178.1 177.3 164.4 177.2 Summer (Ʃ) 86.3 25.0 78.2 22.6 97.3 19.6 118.5 23.0 Water year (Ʃ) 265.2 173.9 261.2 197.4 275.4 196.9 282.9 200.2

The drainage presented an inverse pattern to precipitation, i.e., while higher precipitation volumes were verified in summer (72.9 % in 2014 and 52.1 % in 2015 for the ground-level gauges), higher discharge volumes were recorded in winter. Moreover, the weather data evaluation revealed that summer provided an adequate temperature and solar radiation levels for crop growth, which

87

Chapter 3 consequently increased evapotranspiration and decreased lysimeter outflow (Loos et al., 2007; Harsch et al., 2009). These results are consistent with the data presented by Blume (1992), who reported an annual precipitation drainage rate of 29 % for a greening cover in Gießen, Germany. Additionally, Hermsmeyer (2001) reported a drainage rate ranging from 24 to 39 % over a three-year period in one study with ryegrass and different substrates near Hannover, Germany. In accordance with these previous studies, Scheer (2001) measured a drainage rate ranging from 22 to 44.8 % from April 1998 to April 2000 in Wunstorf, Germany, using different lysimeters and mixtures of recycled aluminum slags and coal combustion residues. In contrast to the aforementioned findings, Harsch et al. (2009) reported higher drainage rates for perennial grasses compared with the lysimeter measurements of this study. Harsch et al. (2009) studied the drainage of three treatments: (1) a grassland mowed three to six times per year, (2) Oak (Quercus robur L.)/Beech (Fagus sylvatica L.) forest and (3) White Pine (Pinus strobus L.) forest, using large-scale lysimeters. These authors observed that the grassland drained 420 mm per year, approximately 53 % of annual precipitation, whereas the Oak/Beech forest percolated 37 % (295 mm) and White Pine 26 % (217 mm) of the annual precipitation, i.e., 791 mm. Harsch et al. (2009) also verified that almost all drainage occurred in winter. Although a coefficient of variation ranging from medium (19.7 % in substrate 2) to high (29.4 % in substrate 1) was found in the drainage between 2014 and 2015, a low variation among the substrates for the outflow in 2014 (CV: 3.6 %) and in 2015 (CV: 6.4 %) was verified. This demonstrates that different proportions of household waste incineration slags and coal combustion residues did not affect the drainage of the substrates employed in this study. Therefore, additional criteria should be taken into account when choosing substrates, such as affordability, availability, stability, and environmental issues.

3.6.2.2 Actual evapotranspiration The annual and biennial actual evapotranspiration of substrates 1, 2, 3 and 4 can be observed in Table 3-5. The mean actual evapotranspiration recorded for substrate 1 was 43.6 mm/month in 2014 and 34.2 mm/month in 2015. The highest monthly actual evapotranspiration was found in summer 2014 (mean of 81.5 mm), followed by summer 2015 (mean of 46.5 mm), whereas the lowest actual evapotranspiration was measured in winter 2014 (5.7 mm) and winter 2015 (mean of 21.9 mm). The total actual evapotranspiration recorded for substrate 1 was 523.0 mm in 2014 (66.4 % of rainfall) and 410.5 mm in 2015 (70.2 % of precipitation). A similar trend was observed for substrates 2, 3 and 4.

88

Chapter 3

Table 3-5: Actual evapotranspiration (mm) according to different substrates at the Heringen experimental site during two hydrological years

Substrate 1 Substrate 2 Substrate 3 Substrate 4

(n = 2) (n = 2) (n = 2) (n = 2) mm mm mm mm Water year / Month 2014 2015 2014 2015 2014 2015 2014 2015 Winter (Ʃ) 34.3 131.3 30.3 105.3 35.2 102.9 48.9 102.9 Summer (Ʃ) 488.7 279.3 496.8 281.7 477.7 284.7 456.5 281.3 Water Year (Ʃ) 523.0 410.5 527.1 387.1 512.9 387.6 505.3 384.2

The results of this study are consistent with the data presented by Blume (1992), who reported an annual evapotranspiration (ETa) value of 490 mm (71 % of precipitation) in the center of Germany, with higher rates in summer (92 %) and lower rates in winter (45 %). The findings of the present study also agree with Mueller et al. (2005), who reported a real evapotranspiration rate ranging from 336 to 505 mm during the vegetation period of perennial ryegrass from April to September in Berlin. The authors highlighted that the variation in water use was correlated with the water table depth, which ranged from 55 to 120 cm, as well as with the weather conditions at the experimental site. Following the same trend as the drainage, the total actual evapotranspiration (mm) showed a medium coefficient of variation (Ø Substrate 1-Substrate 4: 19.4 %) among the evaluation years, however, a low variation was found among the substrates in 2014 (1.9 %) and in 2015 (3.1 %). Figure 3-5 presents the biennial water balance components calculated from 2014 to 2015. Reference crop evapotranspiration (ET0), crop evapotranspiration (ETc) and Haude's evapotranspiration (ETH) are also displayed (FAO, 2014; Haude, 1955).

89

Chapter 3

Substrate 1 Substrate 2 700 0 700 0 600 100 600 100 S1 (mm)S1 500 200 (mm)S2 500 200 - - 400 300 400 300 300 400 300 400 200 500 200 500 100 600 100 600 Precipitation (mm)Precipitation

0 700 0 700 (mm)Precipitation -100 800 -100 800 -200 900 -200 900 Drainage / Evapotranspiration Evapotranspiration Drainage / Drainage / Evapotranspiration Evapotranspiration Drainage / -300 1000 -300 1000 Winter Summer Winter Summer Winter Summer Winter Summer 2014 2015 2014 2015

Time (water years) Time (water years)

Substrate 3 Substrate 4 700 0 700 0 600 100 600 100

500 200 (mm)S4 500 200 S3 (mm)S3 - - 400 300 400 300 300 400 300 400 200 500 200 500 100 600 100 600 Precipitation (mm)Precipitation

0 700 (mm)Precipitation 0 700 -100 800 -100 800 -200 900 -200 900 Drainage / Evapotranspiration Evapotranspiration Drainage / Drainage / Evapotranspiration Evapotranspiration Drainage / -300 1000 -300 1000 Winter Summer Winter Summer Winter Summer Winter Summer 2014 2015 2014 2015

Time (water years) Drainage Precipitation ETa ETo ETc ETH

Figure 3-5: Water balance components of different substrates during winter and summer for 2014 and 2015 (Data are total sum ± standard deviation). Ground-level rain gauges n=4. Drainage n=2. Where ETa: actual evapotranspiration; ET0: reference crop evapotranspiration; ETc: crop evapotranspiration; ETH: Haude´s evapotranspiration The actual evapotranspiration estimated from the lysimeter's water balance was lower than the reference crop evapotranspiration in winter 2014, with an adjusted crop coefficient (Kcadj) of 0.18 for substrate 1, 0.16 for substrate 2, 0.19 for substrate 3 and 0.26 for substrate 4. These crop coefficients are lower than those initial standard values for ryegrass hay or extensive grazing pasture suggested by the FAO (Allen et al., 1998), i.e., 0.95 and 0.30, respectively. This is likely due to the initial establishment of the perennial grasses at the experimental site, from August to September 2013. In contrast, actual evapotranspiration in summer 2014 was similar to the total reference crop evapotranspiration, with an adjusted crop coefficient of 0.95 for substrate 1, 0.96 for substrate 2, 90

Chapter 3

0.93 for substrate 3 and 0.89 for substrate 4. These coefficients are close to those suggested by the FAO (Allen et al., 1998) for midseason (1.05) and late season (1.0) ryegrass hay. Similar values were obtained by Mueller et al. (2005), who reported adjusted crop coefficients for perennial ryegrass in shallow water-table lysimeters that ranged from 0.61 to 0.89. The adjusted crop coefficients in winter 2015 were higher than those measured in winter 2014 because the crops were well established in this season; these values were 0.74 for substrate 1, 0.59 for substrate 2 and 0.58 for substrate 3 and 4. Lower adjusted crop coefficients were obtained in summer 2015 compared with those obtained in summer 2014, i.e., 0.47 for substrate 1, 0.47 for substrate 2, 0.48 for substrate 3 and 0.47 for substrate 4. The summer 2015 adjusted crop coefficients revealed that the water shortage during this period, probably affected the grass growth (Allen et al., 1998; Hillel, 1998).

Overall, ET0, ETc and ETH overestimated the evapotranspiration when compared with actual evapotranspiration (ETa), except in summer 2014. This could be due to the initial crop establishment (verified in winter 2014), less water availability (verified in summer 2015) or any other chemical or physical properties of the substrates, which may have limited the evapotranspiration of the green cover. Slightly higher crop evapotranspiration in relation to actual evapotranspiration was also found by Groh et al. (2015). These authors studied the water balance components of an intensely cultivated ryegrass, “Lolio perennis-Cynosuretum cristati”, considering 4 cuts per year, using six weighing lysimeters (1.5-m depth, 1-m2 area) in Rollesbroich, Germany, during 2014. The authors recorded an actual evapotranspiration (ETa) value of 696.3 mm, equivalent to 65 % of the annual precipitation

(1067 mm), whereas crop evapotranspiration under standard conditions, ETc (Allen et al., 1998), was 698.4 mm.

Nolz et al. (2014) found that the reference crop evapotranspiration (ET0) was greater than the actual evapotranspiration (ETa) in a study carried out using a weighing lysimeter (2.5-m depth, 2.85- m2 area) and irrigated grass in Austria during 2011. The authors reported an annual evapotranspiration (ETa) of 754 mm, which was approximately 96.2 % of total precipitation plus irrigation (784 mm), whereas the reference crop evapotranspiration (ET0) was 826 mm. Nolz et al. (2014) suggested that even when irrigated, the water availability for the grass was sub-optimal.

3.7 Implications and limitations of the study The water balance of the lysimeters revealed that the revegetation procedure on the Heringen potash tailings pile contributed to an actual evapotranspiration of up to 70.2 % of the recorded precipitation during the experiment, which is higher than the evaporation expected for potash tailing piles in Germany, approximately 10 % (K+S KALI GmbH, personal communication, 2015). Higher evapotranspiration rates on potash tailings piles lead to lower brine injection and salt loading 91

Chapter 3 into rivers and streams (Coring and Bäthe, 2011). On the other hand, one needs to take care in extrapolating these results to larger scales because the artificial lower boundary condition of the gravity lysimeters, where drainage occurs when saturation is reached (Klammler and Fank, 2014), does not correspond with the tension head of the potash tailings, which in turn can lead to different upward or downward water fluxes (Hagenau et al., 2015). In addition, surface runoff (Hillel, 1998) and lateral subsurface water flux (Meissner et al., 2008) were not considered in this study. These phenomena probably occur in potash tailings dumps due to their steep slopes (Niessing, 2005). Likewise, higher wind speed and lower temperatures (depending on dump height) can alter the evapotranspiration capacity of a green cover (Chang, 2009; Hillel, 1998; Geiger et al., 2009). Further care must be taken in relation to solar radiation, as the dump can be exposed to different proportions of brightness (Hermsmeyer, 2001). Long-term natural crop diversification (Papke and Schmeisky, 2013) must also be taken into account because different crops have diverse leaf area indexes, heights, root depths, densities and vegetation periods, which can affect water demand (Allen et al., 1998; Klammler and Fank, 2014). Mass reproduction of insects due to mild winter conditions can compromise crop growth, leading to lower evapotranspiration rates (Harsch et al., 2009). Lastly, the quality of the drainage from the substrates must be evaluated to determine whether the substrates are appropriate for use as green covers (Scheer, 2001; Braunisch, 2008; Podlacha, 1999). This study found that a drainage of approximately 30 % of the annual precipitation would take place, and the chemical compounds of this outflow must be considered. Lower seepage rates are predicted when increasing the root depths, crops heights and the rate of fine particles in the substrates (Chapter 4).

3.8 Conclusions In this study, the water balance components of eight non-weighable lysimeters during two hydrological years, 2014 and 2015, were evaluated. Air temperature, soil temperature, air humidity, wind speed and solar radiation were similar between the evaluation years, in contrast to the precipitation which ranged from 788.3 mm in 2014 to 584.4 mm in 2015, based on readings from ground-level gauges. Drainage registered a high variation between the years, albeit a low variation among the substrates. This demonstrates that the different proportions of coal combustion residues and household waste incineration slags led to similar water percolation readings. Therefore, additional criteria should be taken into account when choosing the substrates. The overall total drainage was 271.2 mm in 2014, equivalent to 34.4 % of annual precipitation, and 192.1 mm (32.9 %) in 2015. Likewise, actual crop evapotranspiration was similar between the substrates, totaling 517.1 mm in 2014 (65.6 %) and 392.3 mm (67.1 %) in 2015. An increased evapotranspiration rate on potash 92

Chapter 3 tailings piles decreases brine drainage from these piles and consequently decreases the salt load that reaches rivers and groundwater reservoirs. Considering that the evaporation from the potash tailings piles is approximately 10 %, it is possible to predict that up to 57.1 % of the drainage can be reduced by an evapotranspiration cover. Although when considering the simulations with calibrated Hydrus- 1D model, the evapotranspiration of potash tailings covers may range from 73.3 to 90.1 % according to different root depths and crop heights (Chapter 4, Figure 4-20). The evapotranspiration models, such as the FAO Penman-Monteith reference crop evapotranspiration and Haude's potential evapotranspiration, overestimated the water use of the evapotranspiration cover. As the models assume there is no soil, water or nutrient limitation, it is likely that either the physical chemical characteristics of the substrates or water availability restricted the vegetation growth and consequently, the actual water consumption. Finally, additional characteristics of the drainage, such as chemical properties, should be considered when extrapolating the results of this study to larger scales. Further development of the vegetation e.g. roots and crop heights should be taken into consideration as well.

3.9 References Aboukhaled, A., Alfaro, A., Smith, M., 1982. Lysimeters, FAO Irrigation and Drainage Paper 39, Rome. Allen, R.G., Howell, T.A., Pruitt, W.O., Walter, I.A., Jensen, M.E. (Eds), 1991. Lysimeters for evapotranspiration and environmental measurements. International symposium on lysimetry. New York: American Society of Civil Engineers. Allen, R.G., Pereira, L.S., Howell, T.A., Jensen, M.E., 2011. Evapotranspiration information reporting: I. Factors governing measurement accuracy. Agr. Water Manage. 98, 899-920. Doi: 10.1016/j.agwat.2010.12.015. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration - guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, Rome. Alwast, H., Riemann, A., 2010. Verbesserung der umweltrelevanten Qualitäten von Schlacken aus Abfallverbrennungsanlagen. Umweltbundesamtes. https://www.umweltbundesamt.de/sites/default/files/medien/461/publikationen/4025.pdf (accessed 02 September 2016). Arle, J., Wagner, F., 2013. Effects of anthropogenic salinisation on the ecological status of macroinvertebrate assemblages in the Werra River (, Germany). Hydrobiologia 701, 129-148. Doi: 10.1007/s10750-012-1265-z. Bartsch, M., Fröhlingsdorf, M., 2009. Umwelt: Alarm am Monte Kali. Der Spiegel 35/2009. http://www.spiegel.de/spiegel/print/d-66567967.html (accessed 17 January 2016).

93

Chapter 3

Bäthe, J., Coring, E., 2011. Biological effects of anthropogenic salt-load on the aquatic Fauna. A synthesis of 17 years of biological survey on the rivers Werra and Weser. Limnol. Ecol. Manage. Inland Waters. 41, 125-133. Doi: 10.1016/j.limno.2010.07.005. Bebenburg, P., 2014. K+S: Aus für die Salz-Pipeline. Franfurter Rundschau. http://www.fr- online.de/rhein-main/k-s-aus-fuer-die-salz-pipeline,1472796,28527030.html (accessed 05 September 2016). Beeson, R.C., 2011. Weighing lysimeter systems for quantifying water use and studies of controlled water stress for crops grown in low bulk density substrates. Agr. Water Manage. 98, 967-976. Doi: 10.1016/j.agwat.2011.01.005. Bethune, M.G., Selle, B., Wang, Q.J., 2008. Understanding and predicting deep percolation under surface irrigation. Water Resour. Res. 44, W12430. Doi: 10.1029/2007WR006380. Bilibio, C., Carvalho, J.A., Hensel, O., Richter, U., 2011. Effect of different levels of water deficit on rapeseed (Brassica napus L.) crop. Ciênc. Agrotec. 35, 672-684. Doi: 10.1590/S1413- 70542011000400005. Bilibio, C., Schellert, C., Retz, S., Hensel, O., 2015. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet. Stufe II. Feldversuch auf der Halde IV in Heringen. 2. Zwischenbericht. Teilbericht B (Sickerwasser / Klima). Universität Kassel, Fachgebiet Agrartechnik (unveröffentlichter Bericht). Blombäck, K., Eckersten, H., 1997. Simulated growth and nitrogen dynamics of a perennial rye grass. Agr. Forest Meteorol. 88, 37-45. Doi: 10.1016/S0168-1923(97)00053-1. Blume, H.P., 1992. Handbuch des Bodenschutzes Bodenökologie und -belastung ; vorbeugende und abwehrende Schutzmaßnahmen. (2 ed.). Landsberg/Lech: Ecomed. Blume, H.P., Brümmer, G.W., Fleige, H., Horn, R., Kandeler, E., Kögel-Knabner, I., et al., 2016. Scheffer/Schachtschabel Soil Science. (1 ed.). Berlin: Springer. Blume, H.P., Brümmer, G.W., Horn, R., Kandeler, E., Kögel-Knabner, I., Kretzschmar, R., Stahr,

K., Wilke, B.M., 2010a . Scheffer/Schachtschabel: Lehrbuch der Bodenkunde. (16 ed.). Spektrum Akademischer Verlag. Blume, H.P., Horn, R., Thiele-Bruhn, S., 2010b. Handbuch des Bodenschutzes Bodenökologie und -belastung; vorbeugende und abwehrende Schutzmaßnahmen. (4 ed.). Weinheim: Wiley-VCH. Brady, N.C., Weil, R.R., 2014. The nature and properties of soils. (14 ed.). Harlow: Pearson. Braukmann, U.; Böhme, D., 2011. Salt pollution of the middle and lower sections of the river Werra (Germany) and its impact on benthic macroinvertebrates. Limnologica 41, 113-124. Doi: 10.1016/j.limno.2010.09.003. Braunisch, F., 2008. Untersuchungen zum Aufbau einer funktional optimierten Rekultivierungsschicht auf einer hochbasischen Aschendeponie (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 30/2008, Universität Kassel, Witzenhausen. 94

Chapter 3

Chang, J., 2009. Climate and agriculture: An ecological survey. London: Aldine Transaction. Ciceri, D., Manning, D.A.C., Allanore, A., 2015. Historical and technical developments of potassium resources. Sci. Total Environ. 502, 590-601. Doi: 10.1016/j.scitotenv.2014.09.013. Coring, E., Bäthe, J., 2011. Effects of reduced salt concentrations on plant communities in the River Werra (Germany). Limnol. Ecol. Manage. Inland Waters 41, 134-142. Doi: 10.1016/j.limno.2010.08.004. Couto, M. F.; Peternelli, L. A.; Barbosa, M. H. P. 2013. Classification of the coefficients of variation for sugarcane crops. Ciência Rural 43, 957-961. Doi: 10.1590/S0103-84782013000600003. Crawley, M.J., 2014. Statistics. An introduction using R. (2 ed.). Wiley. Deutsche Wetterdienst: Climate Data Center (CDC). 2016. http://www.dwd.de/EN/climate_environment/cdc/cdc_node.html (accessed 17 January 2016). Deutsches Institut für Normung e.V., 1996. Wasserwesen. Begriffe. Normen. (3. Aufl.). Stand der abgedr. Normen: Oktober 1994. Berlin, Beuth (DIN-Taschenbuch / Hrsg.: DIN, Deutsches Institut für Normung e.V, 211). Dietrich, O., Fahle, M., Seyfarth, M., 2016. Behavior of water balance components at sites with shallow groundwater tables: Possibilities and limitations of their simulation using different ways to control weighable groundwater lysimeters. Agr. Water Manage. 163, 75-89. Doi: 10.1016/j.agwat.2015.09.005. Ehlers, W., Goss, M., 2003. Water dynamics in plant production. Wallingford: CABI Publishing. European Comission, 2012. Report from the Commission to the European Parliament and the Council on the implementation of the Water Framework Directive (2000/60/EC) - River Basin Management Plans. http://eur-lex.europa.eu/legal- content/EN/TXT/PDF/?uri=CELEX:52012DC0670&from=EN (accessed 05 September 2016). European Commission, 2009. Reference document on best available techniques for management of tailings and waste-rock in mining activities. http://eippcb.jrc.ec.europa.eu/reference/BREF/mmr_adopted_0109.pdf (accessed 17 August 2016). European Parliament, 2000. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000. Establishing a framework for community action in the field of water policy. http://data.europa.eu/eli/dir/2000/60/oj (accessed 05 September 2016). Feldhake, C.M., Boyer, D.G., 1986. Effect of soil temperature on evapotranspiration by C3 and C4 grasses. Agr. Forest Meteorol. 37, 309-318. Doi: 10.1016/0168-1923(86)90068-7. Feuerborn, H., 2011. Coal combustion products in Europe - An update on production and utilisation, standardisation and regulation. World of Coal Ash Conference (WOCA), May 9-12, 2011, in Denver, Colorado, USA. http://www.flyash.info/2011/007-feuerborn-2011.pdf (accessed 02 September 2016). 95

Chapter 3

Field, A.P., Miles, J., Field, Z., 2012. Discovering statistics using R. London: SAGE. Fisher, D.K., 2012. Simple weighing lysimeters for measuring evapotranspiration and developing crop coefficients. Int. J. Agric. Biol. Eng. 5, 35-43. Doi: 10.3965/j.ijabe.20120503.004. Fletcher, A.L., Sinclair, T.R., Allen, L.H., 2007. Transpiration responses to vapor pressure deficit in well watered ‘slow-wilting’ and commercial soybean. Environ. Exp. Bot. 61, 145-151. Doi: 10.1016/j.envexpbot.2007.05.004.

Food and Agriculture Organization of the United Nations (FAO), 2014. ET0 calculator. Water Development and Management Unit, Information Resources. http://www.fao.org/land- water/databases-and-software/eto-calculator/en/ (accessed 17 January 2016). Gebler, S., Hendricks Franssen, H.J., Pütz, T., Post, H., Schmidt, M., Vereecken, H., 2015. Actual evapotranspiration and precipitation measured by lysimeters. A comparison with eddy covariance and tipping bucket. Hydrol. Earth Syst. Sci. 19, 2145-2161. Doi: 10.5194/hess-19- 2145-2015. Geiger, R., Aron, R.H., Todhunter, P., 2009. The climate near the ground. (7 ed.). Lanham: Rowman and Littlefield. González-Talice, J., Yuri, J.A., Lepe, V., Hirzel, J., Del Pozo, A., 2012. Water use in three apple cultivars from the second season to sixth season in a drainage lysimeter. Sci. Hortic. 146, 31- 136. Doi: 10.1016/j.scienta.2012.07.034. Groh, J., Pütz, T., Vanderborght, J.; Vereecken, H., 2015. Estimation of evapotranspiration and crop coefficient of an intensively managed grassland ecosystem with lysimeter measurements. Gumpensteiner Lysim. 16, 107-112. Häckel, H., 1999. Meteorologie. (4 ed.). Stuttgart: Ulmer. Hagenau, J., Meissner, R., Borg, H., 2015. Effect of exposure on the water balance of two identical lysimeters. J. Hydrol. 520, 69-74. Doi: 10.1016/j.jhydrol.2014.11.030. Harsch, N., Brandenburg, M., Klemm, O., 2009. Large-scale lysimeter site St. Arnold, Germany. Analysis of 40 years of precipitation, leachate and evapotranspiration. Hydrol. Earth Syst. Sci. 13, 305-317. Doi: 10.5194/hess-13-305-2009. Hart, R.T., 1989. Salt tailings. The ultimate saline reclamation challenge. In: Walker, D.G., Powter, C.B.; Pole, M.W. (Eds.), Reclamation: A global perspective. (1 ed.). Calgary Alberta, Canada, pp. 543-554. Haude, W., 1955. Zur Bestimmung der Verdunstung auf möglichst einfache Weise. Dt. Wetterdienst, 11, 1-24. Hauser, V.L., 2009. Evapotranspiration covers for landfills and waste sites. Boca Raton: CRC Press. Hensel, O., Retz, S., Schellert, C., 2014. Begrünungskonzept für Kalirückstandshalden im Werra- Gebiet. Stufe II. Feldversuch auf der Halde IV in Heringen. 1. Zwischenbericht. Teilbericht B

96

Chapter 3

(Sickerwasser / Klima). Universität Kassel, Fachgebiet Agrartechnik (unveröffentlichter Bericht). Hermsmeyer, D., 2001. Soil physical and hydrological evaluation of aluminum recycling by- product as an infiltration barrier for potash mine tailings (Doctoral Dissertation). Hanover University, Welfengarten. Hermsmeyer, D., Diekmann, R., Van Der Ploeg, R.R., Horton, R., 2002. Physical properties of a soil substitute derived from an aluminum recycling by-product. J. Hazard. Mater. 95, 107-124. Doi: 10.1016/S0304-3894(02)00087-0. Hillel, D., 1998. Environmental soil physics: fundamentals, applications, and environmental consideration. Cambridge: Academic Press. Hunt, W.F., Easton, H.S., 1989. Fifty years of ryegrass research in New Zealand. http://www.grassland.org.nz/publications/nzgrassland_publication_2222.pdf (accessed 21 March 2016). International Potash Institute (IPI), 2016. Production and use of potassium chloride. http://www.ipipotash.org/udocs/Chap-1_potash_production.pdf (accessed 23 August 2016). K+S KALI GmbH, 2016a. Global potash deposits. http://www.k-plus- s.com/en/wissen/rohstoffe/index.html?print¼true (accessed 19 July 2016). K+S KALI GmbH, 2016b. Der Vier-Phasen-Plan. http://www.k-plus-s.com/de/gewaesserschutz/4- phasen.html (accessed 05 September 2016). Kirkham, M.B., 2014. Principles of soil and plant water relations. (2 ed.). Amsterdam: Elsevier. Klammler, G., Fank, J., 2014. Determining water and nitrogen balances for beneficial management practices using lysimeters at Wagna test site (Austria). Sci. Total Environ. 499, 448-462. Doi: 10.1016/j.scitotenv.2014.06.009. Konopka, L., 2015. K+S will den Monte Kali in Heringen begrünen. HNA. http://www.hna.de/lokales/rotenburg-bebra/heringen-ort56535/weissen-berge-werden-gruen- hna-5698680.html (accessed 17 January 2016). Kottek, M., Grieser, J., Beck, C., Rudolf, B., Rubel, F., 2006. World Map of the Köppen-Geiger climate classification updated. Meteorol. Z. 15, 259-263. Doi: 10.1127/0941-2948/2006/0130. Lal, R., Shukla, M., 2004. Principles of soil physics. New York: Taylor and Francis. Lamprecht, C., 2016. Heringen. Climate normal 1961 - 1990. Meteostat. http://www.meteostat.net/heringen-hesse/1961-1990/ (accessed 17 January 2016). Loos, C., Gayler, S., Eckart, P., 2007. Assessment of water balance simulations for large-scale weighing lysimeters. J. Hydrol. 335, 259-270. Doi: 10.1016/j.jhydrol.2006.11.017. López-Urrea, R., Montoro, A., Mañas, F., López-Fuster, P., Fereres, E., 2012. Evapotranspiration and crop coefficients from lysimeter measurements of mature ‘Tempranillo’ wine grapes. Agr. Water Manage. 112, 13-20. Doi: 10.1016/j.agwat.2012.05.009. 97

Chapter 3

Lyons, E.M., Pote, J., DaCosta, M., Huang, B., 2007. Whole-plant carbon relations and root respiration associated with root tolerance to high soil temperature for Agrostis grasses. Environ. Exp. Bot. 59, 307-313. Doi: 10.1016/j.envexpbot.2006.04.002. Manning, D.A.C., 2015. How will minerals feed the world in 2050? P. Geologist. Assoc. 126, 14- 17. Doi: 10.1016/j.pgeola.2014.12.005. Meissner, R., Rupp, H., Seeger, J., Ollesch, G., Gee, G.W., 2010. A comparison of water flux measurements. Passive wick-samplers versus drainage lysimeters. Eur. J. Soil Sci. 61, 609-621. Doi: 10.1111/j.1365-2389.2010.01255.x. Meissner, R., Rupp, H., Seyfarth, M., 2008. Advances in out door lysimeter techniques. Water Air Soil Poll. Focus 8, 217-225. Doi: 10.1007/s11267-007-9166-2. Monteith, J.L., 1977. Climate and the efficiency of crop production in Britain. Philos. Trans. Roy. Soc. B 281, 277-294. Doi: 10.1098/rstb.1977.0140. Mueller, L., Behrendt, A., Schalitz, G., Schindler, U., 2005. Above ground biomass and water use efficiency of crops at shallow water tables in a temperate climate. Agr. Water Manage. 75, 117- 136. Doi: 10.1016/j.agwat.2004.12.006. Niessing, S., 2005. Begrünungsmaßnahmen auf der Rückstandshalde des Kaliwerkes - Sigmundshall in Bokeloh (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 25/2005, Universität Kassel, Witzenhausen. Nolz, R., Cepuder, P., Kammerer, G., 2014. Determining soil water-balance components using an irrigated grass lysimeter in NE Austria. Z. Pflanzenernähr. Bodenk. 177, 237-244. Doi: 10.1002/jpln.201300335. Norris, I.B., 1985. Relationships between growth and measured weather factors among contrasting varieties of Lolium, Dactylis and Festuca species. Grass Forage Sci. 40, 151-159. Doi: 10.1111/j.1365-2494.1985.tb01732.x. Papke, G., Schmeisky, H., 2013. Rekultivierung von Rückstandshalden der Kaliindustrie. Ergebnisse aus langjährigen wissenschaftlichen Begleituntersuchungen der Begrünungsflächen auf der Kalirückstandshalde Sigmundshall in Bokeloh. Ökologie und Umweltsicherung, Bd. 35/2013, Universität Kassel, Witzenhausen. Peel, M.C., Finlayson, B.L., Mcmahon, T.A., 2007. Updated world map of the Köppen-Geiger climate classification. Hydrol. Earth Syst. Sci. 11, 1633-1644. Doi: 10.5194/hess-11-1633-2007. Phogat, V., Skewes, M.A., Cox, J.W.; Alam, J., Grigson, G.; Simunek, J., 2013. Evaluation of water movement and nitrate dynamics in a lysimeter planted with an orange tree. Agr. Water Manage. 127, 74-84. Doi: 10.1016/j.agwat.2013.05.017. Piouceau, J., Panfili, F., Bois, G., Anastase, M., Dufossé, L., Arfi, V., 2014. Actual evapotranspiration and crop coefficients for five species of three-year-old bamboo plants under a tropical climate. Agr. Water Manage. 137, 15-22. Doi: 10.1016/j.agwat.2014.02.004. 98

Chapter 3

Podlacha, G., 1999. Untersuchungen zur Substratandeckung mit geringen Schichtstärken aus Bodenaushub-Wirbelschichtasche-Gemischen und ihrer Begrünung (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 16/1999, Universität Kassel, Witzenhausen. Potchter, O., Goldman, D., Kadish, D., Iluz, D., 2008. The oasis effect in an extremely hot and arid climate. The case of southern Israel. J. Arid Environ. 72, 721-1733. Doi: 10.1016/j.jaridenv.2008.03.004 Rauche, H., 2015. Die Kaliindustrie im 21. Jahrhundert. Stand der Technik bei der Rohstoffgewinnung und der Rohstoffaufbereitung sowie bei der Entsorgung der dabei anfallenden Rückstände. (1. Aufl.). Berlin: Springer. Rawashdeh, R.; Maxwell, P., 2014. Analyzing the world potash industry. Resour. Policy 41, 143- 151. Doi: 10.1016/j.resourpol.2014.05.004. Reid, K.W., Getzlaf, M.N., 2004. Decommissioning planning for Saskatchewan's potash mines. British Columbia Mine Reclamation Symposium. University of British Columbia. Doi: 10.14288/1.0042463. Rose, C.W., 2004. An introduction to the environmental physics of soil, water and watersheds. Cambridge: Cambridge University Press. Scheer, T., 2001. Rekultivierung von Rueckstandshalden der Kaliindustrie. Untersuchungen zur Nutzbarkeit aufbereiteter Salzschlacke der Sekundaeraluminium-Industrie als Rekultivierungsmaterial einer Kali-Rückstandshalde (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 20/2001, Universität Kassel, Witzenhausen. Schmeisky, H., Hofmann, H., 2000. Rekultivierung von Rückstandshalden der Kaliindustrie - Untersuchungen zum Salzaustrag, zur Sukzession sowie Maßnahmen und Erkenntnisse zur Begrünung. Ökologie und Umweltsicherung, 19/2000, Universität Kassel, Witzenhausen. Schmeisky, H., Papke, G., 2012. Begrünungseignung aufbereiteter Restmineralien aus Müllverbrennungsschlacken - Bericht über die Ergebnisse des Gewächshausversuches. Umwelsicherung Schmeisky (unveröffentlichter Bericht). Schmeisky, H., Papke, G., 2013. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet - Stufe II Feldversuch Lysimeterfeld auf der Halde IV in Heringen - 1. Zwischenbericht Teilbericht A. Umweltsicherung Schmeisky (unveröffentlichter Bericht). Schmeisky, H., Podlacha, G., 2000. Natural revegetation of saline waste dumps - drought tolerant specialists and halophytes. Landscape Urban Plan. 51, 159-163. Doi: 10.1016/S0169- 2046(00)00106-7. Schwarz, T., 2016. Climate: Heringen - Climate graph, Temperature graph, Climate table. http://en.climate-data.org/location/202980/ (accessed 17 January 2016).

99

Chapter 3

Searls, J.P., 1992. Potash. In Bureau of Mines. Minerals yearbook 1992. Volume 1, U.S. Bureau of Mines, 1007-1033. http://digital.library.wisc.edu/1711.dl/EcoNatRes.MinYB1992v1 (accessed 02 September 2016). Sinclair, T., Fiscus, E., Wherley, B., Durham, M., Rufty, T., 2007. Atmospheric vapor pressure deficit is critical in predicting growth response of “cool-season” grass Festuca arundinacea to temperature change. Planta 227, 273–276. Doi: 10.1007/s00425-007-0645-5. Soldevilla-Martinez, M., Quemada, M., López-Urrea, R., Muñoz-Carpena, R., Lizaso, J.I., 2014. Soil water balance: Comparing two simulation models of different levels of complexity with lysimeter observations. Agr. Water Manage. 139, 53-63. Doi: 10.1016/j.agwat.2014.03.011. Spliethoff, H., 2010. Power generation from solid fuels. Heidelberg: Springer. van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892-898. van Genuchten, M.Th., Leij, F.J., Yates, S.R., 1991. The RETC code for quantifying the hydraulic functions of unsaturated soils. Environmental Protection Agency Report, 600/2-91/065, U.S. Salinity Laboratory. Werra-Kalibergbau-Museum, 2016. Der Monte Kali: Besondere Bergtour mit Weitblick. http://kalimuseum.heringen.de/index.php?menueid=0andm1=7andartikel=71 (accessed 17 Janyary 2016). Wherley, B.G., Sinclair, T.R., 2009. Differential sensitivity of C3 and C4 turfgrass species to increasing atmospheric vapor pressure deficit. Environ. Exp. Bot. 67, 372-376. Doi: 10.1016/j.envexpbot.2009.07.003. Winter, S., 2016. Gewisses Restrisiko. Der Spiegel 16.06.16, 40-43. http://www.spiegel.de/spiegel/print/d-144314347.html (accessed 02 September 2016). World Meteorological Organization, 2008. Guide to meteorological instruments and methods of observation. Geneva: WMO.

100

Chapter 4

4 Simulation of Evapotranspiration and Drainage on Potash Tailings Covers using Hydrus- 1D Carolina Bilibioa*, Oliver Hensela, Rien van Genuchtenb,c, Daniel Uteaud, Stephan Pethd

a Department of Agricultural and Biosystems Engineering - University of Kassel, Nordbahnhofstraße 1a, D-37213 Witzenhausen, Germany b Center for Environmental Studies, CEA, São Paulo State University, UNESP, Rio Claro, SP, Brazil c Department of Earth Sciences, Utrecht University, Utrecht, Netherlands *Corresponding author ([email protected]) d Department of Soil Science - University of Kassel, Nordbahnhofstraße 1a, D-37213 Witzenhausen, Germany

4.1 Graphical abstract

4.2 Highlights  Hydrus-1D was calibrated to simulate evapotranspiration and drainage on potash tailings covers.  The calibrated model improved the agreement between observed and predicted data.  Lower seepage rates are expected when the substrates’ fine fraction is increased.  The hydraulic properties of the evapotranspiration covers showed similar characteristics.

101

Chapter 4

4.3 Abstract Evapotranspiration is the evaporation of soils and the transpiration from crops. It is a crucial part of the water cycle and can be measured using lysimeters. However, lysimeters are generally not designed for long term evaluations due to high maintenance and operation costs and thus evapotranspiration models should be calibrated for making predictions. Considering this, the aim of the present study was to calibrate the Hydrus-1D mathematical model to determine the water balance components of evapotranspiration covers (in this case a thin layer cover) for potash tailings piles. For this, the meteorological data from a lysimeter experimental site located in Heringen, Germany, was used. This experiment was installed in 2013 at the Wintershal plant, a part of the Werra combined potash plant from K+S KALI GmbH. Four technogenic substrates were placed in 8 non-weighable lysimeters. The substrates were made of different proportions of household waste incineration slags and coal combustion residues. The actual evapotranspiration in the field was estimated using the simplified water balance equation (ET = P - D) from 2014 to 2016. Hydraulic properties of the substrates, pH and electrical conductivity were measured in 2014 and in 2016. Validation of the model was made using 27-years of daily data from a weather station located 20 km away. Further simulations were performed using different rates of fine fractions, soil textures and crop parameters. A high association between the calibrated and observed seepage of the substrates was observed, with a variation of 2.9 % or circa 13.6 mm. Moreover, a drainage rate of 24.7 % and an evapotranspiration rate of 75.3 % were found when studying the historical weather data. Lower seepage was estimated when increasing the root depth, crop height and the rate of fine particles, < 2 mm diameter, in the substrates. The substrates showed similar water fluxes and properties as well as a mean pH of 8.5 and an electrical conductivity of 3.0 mS/cm. Keywords Water consumption; Water fluxes; Drainage; Lysimeters; Technogenic substrates

102

Chapter 4

4.4 Introduction Brine drainage from potash tailings piles impacts natural ecosystems due to the high concentration of sodium chloride (Canedo-Argüelles et al., 2017). This drainage may decrease using natural processes, such as evapotranspiration (Rauche, 2015). Evapotranspiration is the second major component in the water cycle, after precipitation (Lamb, 2015; Novák, 2012; Abtew and Melesse, 2013; Tukimat et al., 2012). Researchers have estimated that 60 % of continental precipitation is transported to the atmosphere by evapotranspiration (Lamb, 2015; Novák, 2012). Evapotranspiration is the evaporation of water from the soil and transpiration from the crops (Lamb, 2015). Evapotranspiration is important as it consumes energy, i.e., 2450 joules per gram of water at 20 oC (Novák, 2012); regulates the temperature of leaves and soil (Thornthwaite, 1948); determines the water requirements for crops and irrigation schedules (Goyal and Harmsen, 2014); and recharges surface or ground water (Tukimat et al., 2012). Evapotranspiration also characterizes the climate. For instance, it is humid when precipitation exceeds a certain amount of potential evapotranspiration and arid if rainfall is lower (Thornthwaite, 1948). Additionally, the selection of crops and production regions are also based on evapotranspiration studies (Chang, 2009; Goyal and Harmsen, 2014). Evapotranspiration rates are governed by weather conditions, such as solar radiation, temperature, wind speed, the gradient of water vapor (Allen et al., 1998); available moisture at the root zone; land management (Thornthwaite, 1948); plant type and growth stage (Doorenbos and Pruitt, 1977). Three main concepts describe evapotranspiration: reference evapotranspiration, crop evapotranspiration and actual evapotranspiration (Goyal and Harmsen, 2014; Allen et al., 1998).

Reference evapotranspiration (ET0) refers to the water consumption from a hypothetical grass, with uniform height (0.12 m) and not short in water and nutrients (Allen et al., 1998). Crop evapotranspiration (ETc) incorporates the evaporation and transpiration from a field crop under optimum conditions (Allen et al., 1998). Whereas actual evapotranspiration (ETa) indicates the quantity of water removed from the crops under non-standard environments (Allen et al., 1998). Actual evapotranspiration can be measured using lysimeters (Goyal and Harmsen, 2014; Allen et al., 1998). Lysimeters are containers filled-in with soil or soil monolith (undisturbed soil structure) and placed in the field (Aboukhaled et al., 1982). Lysimeters can have different shapes, i.e., circular, square or rectangular; and be made of diverse materials, such as steel, concrete and plastic (Goyal and Harmsen, 2014). They must be deep enough to allow complete root growth, be surrounded by the studied crops and have a representative area (Goyal and Harmsen, 2014). Lysimeters evaluate evapotranspiration by measuring water balance components, i.e., precipitation, surface runoff, seepage and the variation in the soil-water storage (Lamb, 2015; 103

Chapter 4

Novák, 2012; Abtew and Melesse, 2013). Nevertheless, lysimeter measurements represent a specific area and are not designed for long term measurements due to the high operation and maintenance costs as well as being time consuming (Abtew and Melesse, 2013). Hence, observational data should be used to calibrate evapotranspiration models allowing one to make predictions (Goyal and Harmsen, 2014; Radcliffe and Simunek, 2010). Several studies have simulated water flow processes using the Hydrus mathematical model (Galleguillos et al., 2017; Kodešová et al., 2014; Li et al., 2014; Tan et al., 2014). Hydrus simulates infiltration, evaporation, transpiration, redistribution and discharge water through saturated and unsaturated mediums using Richards equation (Simunek et al., 2008; Radcliffe and Simunek, 2010; Simunek et al., 2013). Heat and solute flow can also be considered (Radcliffe and Simunek, 2010; Lamb, 2015). However, few studies have been conducted to simulate the water fluxes from evapotranspiration covers for potash tailings piles using field experimental data. Therefore, the aim of this study was to calibrate and validate the Hydrus-1D model to make predictions using observational data from a lysimeter experiment. The lysimeter experiment was installed at the Wintershal plant, from K+S KALI GmbH, in July 2013. There were 4 treatments (technogenic substrates), with two repetitions. The substrates were made up of different proportions of household waste incineration slags and coal combustion residues (Bilibio et al., 2017). The observational data included meteorological conditions, hydraulic properties, pH, electrical conductivity and drainage from the substrates. The observed drainage refers to the water volume collected on the bottom of the lysimeters. Five calibration types were performed to find the best fitting parameters for the Hydrus-1D validation. The validation was done using daily precipitation, air temperature, solar radiation, wind speed and relative air humidity from a weather station located circa 20 km away. Lastly, predictions of evapotranspiration and drainage were made using different rates of fine fractions, soil texture and crop parameters, such as root depth and crop height. The results from this investigation can contribute to improving the efficiency of evapotranspiration covers for potash tailings piles.

4.5 Material and methods The next items of the article discuss the location of the experimental site, design of the treatments, meteorological data and the methods to estimate the hydraulic properties, pH and electrical conductivity of the substrates. Additionally, the configuration of the Hydrus-1D to make the calibration, validation and predictions of water fluxes are shown.

104

Chapter 4

4.5.1 Experimental site and design The experiment was conducted on the potash tailings pile from the Wintershall potash plant, located at 50° 53' 160'' North and 9° 59' 12'' East, on the outskirts of the Heringen city, Germany, Figure 4-1.

Figure 4-1: Aerial view of potash tailings from Wintershall potash plant on 14 May 2012. Size of the picture: 1500 pixels in horizontal versus 1125 pixels in vertical image. The horizontal ground sampling distance, GSD, is 1.4462 m/pixel and the vertical GSD is 1.4460 m/pixel (TerraServer, 2016) The experiment had 4 treatments with two repetitions, consisting of eight non-weighable lysimeters, Figure 4-2. The lysimeters were 3-m deep and covered an area of 2 m². The treatments comprised four substrates made of different proportions of household waste incineration slags (0- 12 mm sieve) and coal combustion residues. Substrate 1: 80 % household waste incineration slags; 20 % coal combustion residues. Substrate 2: 70 % household waste incineration slags; 30 % coal combustion residues. Substrate 3: 60 % household waste incineration slags; 10 % of washed sand from gravel extraction; 30 % coal combustion residues. Substrate 4: 50 % household waste incineration slags; 30 % coal combustion residues; 10 % furnace bottom ashes with particle sizes between 0.2 and 2 mm, labelled “Kesselsand”; 10% original bottom ashes with particle sizes from 0 to 6.3 mm, labelled “Feinasche”. A filter layer of 0.40 m was placed at the bottom of the lysimeters to avoid washing out the substrates and breaking the capillary fringe on the lower boundary of the substrates (Seiler and Gat, 2007). Sand and different diameters of gravel were used, fine medium and coarse gravels.

105

Chapter 4

Figure 4-2: Distribution of the lysimeters and treatments at the experiment field

A seed mixture containing 65 % perennial ryegrass (Lolium perenne L.), 25 % red fescue (Festuca rubra L.) and 10 % Kentucky bluegrass (Poa pratensis L.) was used from 5 August to 26 September 2013, totaling 70 g/m2 (Schmeisky and Papke, 2013). In addition, the annual amount of fertilizer was 83 g/m2 in 2013, 193 g/m2 in 2014, 94 g/m2 in 2015, and 158 g/m2 in 2016, consisting of 61 g/m2 of nitrogen, 80 g/m2 of phosphorus, 79 g/m2 of potassium and 9 g/m2 of magnesium (Schmeisky and Papke, 2013; Papke and Schmeisky, 2017).

4.5.2 Meteorological data Micrometeorological parameters were registered automatically by a Thies-Clima weather station, equipped with a Datalogger DLx-MET. Wind speed (m/s, 3-m height), air temperature (2- m height), soil temperature (0.3-m depth), relative air humidity (2-m height) and solar radiation (2- m height) were recorded at 10-min intervals. Precipitation was assessed by the Thies weather station using a tipping bucket system; 4 rain gauges installed at ground level; and 5 gauges installed at 1-meter height. Precipitation in gauges at 1-meter-high and on ground-level were measured on Thursdays, circa 10:00 am. Due to technical problems with the Thies weather station from 05.08.2016 to 18.08.2016, the weather measurements available for Eichhof (Bad Hersfeld) were incorporated for this interval. Eichhof (Bad Hersfeld) is located at 50o 50' 51.7'' North and 9° 41' 8.1'' East, and at circa 202 m altitude (Landesbetrieb Landwirtschaft Hessen, 2017).

106

Chapter 4

The weather sensors in Eichhof were placed at similar heights from the ground as the ones in Heringen (Werra), except for wind speed (2.5 m height) (Landesbetrieb Landwirtschaft Hessen, 2017). Therefore, for further studies the wind speed in Heringen (Werra) and in Eichhof (Bad Hersfeld) was adjusted to 2.0 m height. Before the incorporation of weather data in 2016, Spearman correlation studies were performed between the Eichhof (Bad Hersfeld) and Heringen (Werra) daily weather data from 01.11.2013 to 31.10.2015. For solar radiation, a correlation coefficient of 0.94 (p = 0) was found, whereas a correlation coefficient of 0.93 (p = 0) was found for the minimum air temperature; 0.99 (p = 0) for maximum air temperature; 0.98 (p = 0) for soil temperature; 0.83 (p = 0) for wind speed; and 0.88 (p = 0) for the relative air humidity. The Spearman correlation coefficients show the strength of association between two variables in nature (Field, 2013). These coefficients are classified as very weak (0.0 to 0.19), weak (0.20 to 0.39), moderate (0.40 to 0.69), strong (0.70 to 0.89) and very strong (0.90 to 1.0) (Ludwig, 2015). Table 4-1 shows the meteorological parameters from the Heringen experimental site. Potential evapotranspiration estimated according to Haude´s method, non-standard reference evapotranspiration and crop evapotranspiration assessed according to FAO´s method (Allen et al., 1998) are also presented.

Table 4-1: Meteorological data of the Heringen experimental field during three hydrological years

Parameter Unity 2014 2015 2016 Mean SD CV Precipitation ground-level gauges (Ʃ) mm 788.3 543.8 683.9 672.0 122.7 18.3 Precipitation Thies weather station (Ʃ) mm 778.8 508.1 - - - - Precipitation 1-m high gauges(Ʃ) mm 710.0 484.9 603.4 599.4 112.6 18.8 Minimum air temperature (Ø) oC 6.8 5.8 6.7 6.4 0.6 8.6 Maximum air temperature (Ø) oC 13.8 13.0 13.7 13.5 0.4 3.2 Mean air temperature (Ø) oC 10.0 9.2 10.0 9.7 0.5 4.7 Mean substrate temperature (Ø) oC 11.0 10.2 10.7 10.6 0.4 3.8 Relative air humidity (Ø) % 82.4 80.0 81.6 81.3 1.2 1.5 Wind speed / 2-m height (Ø) m/s 2.5 2.9 2.7 2.7 0.2 7.4 Solar radiation (Ø) W/m² 118.9 127.0 122.9 122.9 4.1 3.3 ET(H) (Ʃ) mm 508.9 571.7 503.2 527.9 38.0 7.2 ET0-n/ref (Ʃ daily) mm 647.5 721.4 675.8 681.6 37.3 5.5 ETc (Ʃ) mm 670.8 750.9 703.1 708.3 40.3 5.7

ET(H): Haude´s potential evapotranspiration; ET0: Reference evapotranspiration; ETc: Crop evapotranspiration; SD: standard deviation Figure 4-3 provides an overview of the daily variation of minimum and maximum air temperature, relative air humidity, solar radiation and wind speed registered at the experimental site during three hydrological years.

107

Chapter 4

40 2014 2015 2016 2014 2015 2016 100 30 C) o 20 80

10 60 0 40 Air temperature ( Air temperature -10 Relative air air (%) humidityRelative

-20 20 1-Feb-2014 1-Feb-2015 1-Feb-2016 1-Feb-2014 1-Feb-2015 1-Feb-2016 1-Nov-2013 1-Aug-2014 1-Nov-2014 1-Aug-2015 1-Nov-2015 1-Aug-2016 1-May-2014 1-May-2015 1-May-2016 1-Nov-2013 1-Aug-2014 1-Nov-2014 1-Aug-2015 1-Nov-2015 1-Aug-2016 1-May-2014 1-May-2015 1-May-2016 Time (days) Time (days)

400 10 2014 2015 2016 2014 2015 2016 350 8 300

250 6 200 4 150

100 spedd Wind(m/s)

Solar radiation (W/m²) Solar 2 50

0 0 1-Feb-2014 1-Feb-2015 1-Feb-2016 1-Feb-2014 1-Feb-2015 1-Feb-2016 1-Nov-2013 1-Aug-2014 1-Nov-2014 1-Aug-2015 1-Nov-2015 1-Aug-2016 1-May-2014 1-May-2015 1-May-2016 1-Nov-2013 1-Aug-2014 1-Nov-2014 1-Aug-2015 1-Nov-2015 1-Aug-2016 1-May-2014 1-May-2015 1-May-2016 Time (days) Time (days)

Figure 4-3: Daily minimum (gray line) and maximum (black line) air temperature, relative air humidity, solar radiation and wind speed (2-m height) registered at the experimental site during three hydrological years The highest precipitation values were found in the ground-level gauges, which ranged from 543.8 mm in 2015 to 788.3 mm in 2014. The mean ground-level precipitation registered from 2014 to 2016, 672 mm, was similar to the historical precipitation in Heringen (Werra) from 1961 to 1990, which is 684 mm per year (Deutsche Wetterdienst, 2017a). On average, ground level gauges registered 12.1 % more precipitation compared with 1-m high gauges. This may be caused by wind-induced losses of precipitation in 1-m high gauges, which has been extensively reported in the literature (Richter, 1995; Braunisch, 2008; Podlacha, 1999; Niessing, 2005). Lower differences were found among the precipitation measured in ground-level gauges and the precipitation recorded by the Thies-weather station, consisting of 1.2 % in 2014 and 108

Chapter 4

7.0 % in 2015. A medium coefficient of variation among the years for precipitation was found for the ground-level gauges, 18.3 %, and for 1-m high gauges, 18.8 %. The mean temperature recorded at the experimental site, 9.7 oC, is slightly higher than the historical mean for Heringen (Werra), 8.4 oC (Lamprecht, 2017), although the minimum of 6.4 oC and the maximum of 13.5 oC was recorded. Solar radiation ranged from 118.9 W/m² in 2014 to 127.0 W/m² in 2015, which is consistent with the mean in Germany, from 110 to 135 W/m². Figure 4-3 shows the seasonal variation of solar radiation, i.e., higher values in summer months (May- October, mean 175.8 W/m²) and lower ones in winter months (November-April, mean 70.1 W/m²). Relative air humidity presented a mean value of 81.3 % with very low variation among the water years, 1.5 %. Lower mean relative air humidity was verified in summer and higher in winter, 76.5 % and 86.1 % respectively. For the potential evapotranspiration using the Haude´s method, a mean value of 527.9 mm/year was verified and for the non-standard reference evapotranspiration a mean value of 681.6 mm/year was estimated. A non-standard ET0 was used because the Thies-weather station was situated above a non-standard field. Standard measurements to estimate the Penman–Monteith reference evapotranspiration are made above an extensive grass area and not short in water (Allen et al., 1998; Pereira et al., 2015). Higher temperatures can be measured in non-standard conditions, which consequently may overestimate the ET0 (Allen et al., 1998). The mean crop evapotranspiration, 708.3 mm/year, was assessed using a crop coefficient of 0.95 for the initial stage (from January to March); 1.05 for the mid-season (from April to October); and 1.0 for the late season (from November/December) (Allen et al., 1998; Hoffman et al., 2007).

4.5.3 Drainage and evapotranspiration assessment Discharge lines connected to the lysimeters drained percolated water. These lines were linked to 60-L barrels placed in a shelter nearby. The amount of drained water was first recorded on 26 July 2013 and was then recorded weekly, on Thursdays between 9 a.m. and 10 a.m. The water balance was determined considering the water entering and leaving the system (Hauser, 2009; Aboukhaled et al., 1982; Abtew and Melesse, 2013):

푃 + 퐼 ± 푅표 = 퐸푇 + 푃푅퐾 + 퐿 ± ∆푆푊 + 푒푟푟표푟 (4-1) Where P is the precipitation; I is the irrigation; Ro is the surface runoff; ET is the actual evapotranspiration; PRK is the deep percolation; L is the lateral flow; ΔSW is the change in soil water storage; error is the lack of balance in the measured terms. As the present study was performed within a confined system, non-weighable lysimeters, the surface runoff and lateral fluxes were not studied (Aboukhaled et al., 1982). Additionally, the water

109

Chapter 4 storage change (± ΔSW) can be considered constant at the beginning and at the end of a water year (Novák, 2012; Seiler and Gat, 2007). Hence, the actual evapotranspiration was determined as residual term of the simplified water balance expression (Bethune et al., 2008):

퐸푇푎 = 푃 − 퐷 (4-2)

Where ETa is the actual evapotranspiration (mm), P is the ground-level precipitation (mm), and D is the lysimeters drainage (mm).

4.5.4 Hydraulic properties, pH and electrical conductivity of the substrates Hydraulic properties, pH and electrical conductivity of the substrates were studied with disturbed and undisturbed samples collected in 2014 and in 2016.

4.5.4.1 Disturbed samples Disturbed samples were collected with a Dutch auger at 0.0-0.2; 0.2-0.4; 1.4-1.6; 2.0-2.2; 2.4- 2.6; 2.8-3.0 m depths surrounding the lysimeters according to each substrate. In 2014 two replicates were considered, totaling 12 samples per treatment. However, in 2016 3 repetitions were performed, totaling 18 samples per substrate. After removing the samples from the field, the samples were placed in plastic bags and taken to the laboratory, where the pH, electrical conductivity, particle size distribution, wettability and color were evaluated. The substrates’ pH was determined based on a practice from the German Institute for Standardization DIN ISO 10390 (2005). For this, 20 grams of each air dried and sieved substrate sample was first placed in plastic containers. Subsequently, 100 ml of calcium chloride solution (0.01 mol/l) was added to each numbered vessel. Then, the containers were stirred for 60 minutes in a Kottermann electric stirrer to obtain homogenized suspension. Afterwards, the pH electrode (SI Analytics pH-Electrode BlueLine 26 pH NTC 10; Sartorius portable meter PT-10) was dipped in the solution for pH reading. To determine the electrical conductivity, the method suggested by the DIN ISO 11265 (1997) was followed. For this, 20 grams of each air-dried and sieved sample and 100 ml of distilled water were added to numbered containers. Then, the plastic vessels were stirred for 30 minutes. Later, the suspension was filtered through a folded filter paper (Schleicher & Schüll 595 1/2, retention range of 4 - 7 µm) and the electrical conductivity was evaluated using a conductivity meter (WTW Pocket Meter Multi 340i). The contact angle of substrate 1 and 4 at 0.20-0.40 m depth was studied using the DataPhysics DCAT 11 device. It was performed 8 repetitions with each substrate. A contact angle of zero degrees was found, which means that the substrates have no wetting restrictions (Blume et al., 2016).

110

Chapter 4

The particle size distribution of the substrates was determined according to the German Institute for Standardization DIN ISO 11277 (2002) by sieving sand-size particles and sedimentation in the water from silt and clay-size particles according to Stokes' Law (Blume et al., 2016). In addition, the coarse fraction of the substrates was determined, which comprises particles greater than 2 mm (Blume et al., 2016). For this, first 30 grams of each sample were placed in a 1-liter glass beaker.

Then, hydrogen peroxide (H2O2, 30% volume concentration) was added to oxidize the organic matter. When there was no reaction between the H2O2 and the substrates, the samples were oven dried for 48 hrs at 40 oC. After that a solution was prepared with 20 grams of each substrate sample, 200 ml of distilled water and 75 ml of dispersing agent (tetra-Sodium diphosphate decahydrate) in 300 ml plastic bottles. Next, the plastic bottles were horizontally placed in a mechanical shaker (GFL 3018) for circa 20 h at 210 shakes/minute. Afterwards, the dispersed suspension was passed through a 2 mm sieve to separate the coarse fraction (> 2 mm diameter) from the fine fraction (< 2 mm diameter). With the fine dispersed fraction, the sieving was proceeded to obtain the coarse sand-size particles (from 630 to 2000 µm), medium sand-size particles (from 200 to 630 µm) and fine sand-size particles (from 63 to 200 µm). The resulting 1 liter dispersed solution was used in the sedimentation process to measure the silt and clay-size particle rates. For this, the sedimentation glass tubes were stirred for 1 min. Then 10 ml of the dispersed solution was sucked with a glass pipette after the sedimentation time, which varied depending on the room temperature (21-23 oC). Soon after, the solution samples were placed in glass vessels and dried at 105 oC for 24 h until they reached a constant weight. From the grain size distribution, the coefficient of uniformity and the the median equivalent diameter, d50, from the substrates were estimated. The coefficient of uniformity is assessed by: 푑 (4-3) 푈 = 60 푑10

Where U is the uniformity coefficient, dimensionless; d60 and d10 are the sieve diameters (mm) where 60 % and 10 % of the grains passed through the sieves. The median equivalent diameter, d50, referes to the sieve diameter which separates the grain mass in two parts, 50 % above and 50 % below the speficific diameter (Horton et al., 2016). The color of the substrates was evaluated in the laboratory using the Munsell soil color charts (Munsell Color Company, 1994), disturbed and undisturbed samples from 0.0 - 0.64 m depth.

4.5.4.2 Undisturbed samples In 2014, vertical undisturbed samples were collected from substrates 1 and 4 using stainless steel rings with cutting edges on the lower part from 0.0 to 0.6 m depth. Three repetitions were completed at each depth, totaling 30 samples.

111

Chapter 4

In 2016, a second horizontal undisturbed sampling was carried out at 0.0-0.04, 0.20-0.24, 0.40- 0.44, 0.60-0.64 m depths considering the four different substrates (S1-S4). 12 repetitions at each depth were collected, totaling 192 samples. At the experimental site, the samples were placed in plastic containers and covered with plastic films to minimize evaporation. In the laboratory, the samples were maintained in a room at circa 12 degrees to proceed with bulk density, particle density, water retention curve and saturated hydraulic conductivity measurements. Dry bulk density was estimated by the ratio of substrate dry mass and sample volume whereas the particle density of substrates 1 and 4 at 0.20-0.26 and 0.40-0.46 m depth (sampling from 2014) was performed using the pycnometer method (DIN 18124, 2011).

The graphical representation of water content (ϴ) versus matric potential (ψm) was studied using sand-bath and suction ceramics plates at 5 hPa; 10 hPa; 20 hPa; 30 hPa; 60 hPa; 150 hPa; 300 hPa and 500 hPa. Pressure chambers were used to evaluate the water content versus matric potential of the undisturbed samples at 15000 hPa. The water retention curve was adjusted according to the van Genuchten model, Eq. 4-4 (van Genuchten, 1980) and the soil water retention parameters were obtained using RETention Curve software (RETC, van Genuchten et al., 1991).

휃푠 − 휃푟 (4-4) 휃(휓푚) = 휃푟 + ( 푛 푚) , 휓푚 < 0 ⌈1 + (훼 . 휓푚) ⌉

Where Ɵr is the residual water content (cm³/cm³), representing the water content where hydraulic conductivity approximates zero (Radcliffe and Simunek, 2010); Ɵs is the water content at soil saturation (cm³/cm³), also considered the total porosity if there is no entrapped air (Radcliffe and

Simunek, 2010); ψm is the matric potential (hPa); α is related to the air-entry at the saturated zone and is equal to the inverse dimension of matric potential (1/cm) (Radcliffe and Simunek, 2010); n is associated with the steepness (slope) of the water retention curve and presents a fixed relation with m, both dimensionless (van Genuchten, 1980), in which: 1 (4-5) 푚 = 1 − ; 푛 > 1 푛 The moisture content of 15000 hPa from substrate 3 at 0.40-0.64 m deep, was adjusted according to the upper layers (0.0-0.24 m depth) because the values obtained during the measurement, 0.18 cm³/cm³, did not fit the van Genuchten model.

Using field capacity 60 hPa; permanent wilting point 15000 hPa; total pore volume ≅ 휃푠, the moisture volume (%) released by gravity was estimated, as well as the plant availability and unavailable water (Radcliffe and Simunek, 2010). The water volume released by gravity is the difference between saturation and field capacity. However, available plant moisture from medium pores is the difference between field capacity and permanent wilting point. Plant unavailable water

112

Chapter 4 refers to the moisture at the permanent wilting point, or matric potential of -15000 hPa (Blume et al., 2016). The soil-water retention curve also allowed to study the pore size distribution. The large coarse pores have an equivalent diameter of >50 µm (< 60 hPa). Yet the tight coarse pores present an equivalent diameter from 50 to >10 µm (60 - 300 hPa), medium pores from 10 to > 0.2 µm (300 - < 15000 hPa), and fine pores ≤ 0.2 µm (≥ 15000 hPa) (AG Boden, 2005). The saturated hydraulic conductivity of substrates 1-4 was measured using a falling head permeameter (Eijkelkamp 09.03 Hauben water permeameter) and followed the recommendations of the German Institute for Standardization DIN 19683-9 (2012). After measuring the saturated hydraulic conductivity, the dry bulk density and the fine and coarse fraction were estimated using a sieve shaker (Eijkelkamp VS 1000) and a set of sieves at 2 and 6.3 mm diameter. The results of the saturated hydraulic conductivity measured in the laboratory using undisturbed samples were compared with the saturated hydraulic conductivity of substrates measured in the field using a double ring infiltrometer. The test was performed for 60 minutes. Based on the cumulative infiltration one could estimate the saturated hydraulic conductivity using A and N parameters of the Kostiakov's infiltration model (Silva, 2005):

퐼 = 퐴 푇푁 (4-6)

푁 − 1 (4-7) −0.001 푁 − 2 𝑖푏 = 퐴 푁 ( ) 퐴 푁 (푁 − 1) Where I is the cumulative infiltration capacity (mm), T is the time (min), A and N are constants with A > 0 and 0 < N < 1, ib is the basic infiltrability (mm/min). As the infiltration test was performed with 0.10 m hydraulic head, the basic infiltrability of the substrates was corrected by the following expression (Simões et al., 2005):

푓푐 = 1.29 퐻−0.29 (4-8) Where fc is the correction factor of the basic infiltrability, H is the hydraulic load (cm). At last, the volumetric water content of the substrates was assessed weekly, on Thursdays, from 25.06.2015 to 27.10.2016 (71 measurements) using a moisture meter sensor, ML3 ThetaProbe. This device comprises 4 needles, 6 cm long connected with a cable to a portable datalog. Circa 20 repetitions per substrate were performed. The device was calibrated between October and December 2015 according to the instructions provided by the manufacturer (Delta-T Devices, 2013).

4.5.5 Simulations and calibration of Hydrus-1D The direct simulation of the water fluxes from substrates 1-4 was performed with the experimental observations from the first two water years. The water year 2016 was not considered due to the lack of precipitation data from the Thies weather station. Moreover, no correction was 113

Chapter 4 conducted in the precipitation from the weather station because the differences with ground-level gauges were low. As mentioned above, a low coefficient of variation among the ground level gauges and the Thies weather station was found, comprising 1.2 % in 2014 and of 7.0 % in 2015. The main inputs and settings of Hydrus-1D, version 4.16.0110 (Simunek et al., 2014), are presented in Table 4-2.

Table 4-2: Hydrus 1-D inputs for the water balance simulation

Parameters and variables Inputs Main processes Water flow, root water uptake Length units cm Type of flow Vertical plane Number of materials 4 set of hydraulic parameters per substrate Number of layers (mass balance sub-regions) 5 (0.0-0.19; 0.20-0.39; 0.40-0.59; 0.60-1.99, 2.0 - 2.6 m depth) Depth of the profile (cm) 260 Time units Days Initial time 0 (01.11.2013) Final time 730 (31.10.2015) Initial time step 0.001 (default) Minimum time step 1e-005 (default) Maximum time step 1 Print times 4 Number of time-variable boundary records 730 Number of meteorological records 730 Potential evapotranspiration Penman-Monteith equation Maximum number of iterations 10 (default) Water content tolerance 0.001 (default) Pressure head tolerance 1 (default) Lower optimal iteration range 3 (default) Upper optimal iteration range 7 (default) Lower time step multiplication factor 1.3 (default) Upper time step multiplication factor 0.7 (default) Lower limit of the tension interval 1 e-006 (default) Upper limit of the tension interval 10000 (default) Hydraulic model van Genuchten-Mualem Hysteresis No Initial condition In pressure head Tortuosity parameter in the conductivity function 0.5 Upper boundary condition Atmospheric boundary condition with surface runoff Lower boundary condition Seepage face, h=0 Water uptake reduction model Feddes´s parameters for grass Solute stress No solute stress Geographical and Meteorological parameters Solar radiation, Latitude 50 degrees North, Altitude 409 m Crop data Growth data in tables. Leaf area index estimated from crop height of alfalfa and other field crops.

114

Chapter 4

Profile discretization (split) 261 nodes (interfaces) or 260 uniform cells with 1 cm deep Root distribution Decreasing linear distribution from 1 at the second node (1 cm depth) to 0 at 18 cm depth Initial pressure head Increasing linear distribution from -100 hPa at top nodes to 0 hPa at the bottom nodes Depth of the observation nodes (cm) 5 cm, 10 cm; 20 cm, 40 cm, 60 cm, 120 cm, 200 cm, 240 cm, and 260 cm

The water flow and root water uptake were simulated. Water flow was selected because there is a transient flow in the lysimeters, which means the water fluxes differ in time and depth during the evaluation time (Simunek et al., 2014; Radcliffe and Simunek, 2010). The root water uptake represents the sink term included in Richard´s equations for one dimensional water flux (Simunek et al., 2013). 휕휃 휕 휕ℎ (4-9) = [퐾 ( + 푐표푠훼)] − 푆(ℎ) 휕푡 휕푥 휕푥 Where h is the pressure head (cm); x is the spatial coordinate (cm); Ɵ is the volumetric water content (cm³/cm³); t is time (days); S is the sink term (cm³/cm³/day); α is the angle between the flow direction and the vertical axes, designated zero degrees for vertical water flow (cos 0o = 1); K is the unsaturated hydraulic conductivity function (cm/day) (Simunek et al., 2013). Four materials were used to simulate the water flow within the lysimeters because 4 sets of hydraulic properties per substrate were measured. The hydraulic parameters of the substrates were adjusted according to the van Genuchten-Mualem model, Eq. 4-4, 4-5 and 4-10 (Simunek et al., 2013).

푚 2 푙 1/푚 (4-10) 퐾(ℎ) = 퐾푠 푆푒 [1 − (1 − 푆푒 ) ]

Where K(h) is the unsaturated hydraulic conductivity function; Se is the effective saturation

(dimensionless); Ks is the saturated hydraulic conductivity (cm/day); and l is the pore-connectivity parameter, which is 0.5 for most soils (Simunek et al., 2013; Mualem, 1976). The potential evapotranspiration was estimated using the Penman-Monteith equation (Eq. 4-11) according to the meteorological parameters measured with the Thies weather station. Moreover, the potential evapotranspiration considered a root depth of 10 cm in October 2014 and 18 cm in October 2015 (Papke and Schmeisky, 2017). The crop height was set to 30 cm during the two water years. Perennial ryegrass and red fescue grow from 30 to 100 cm height according to the soil fertility and water availability (Hannaway et al., 1999; St. John et al., 2012) whereas Kentucky bluegrass grows up to 60 cm height (Bush, 2002). Leaf area index in Hydrus-1D was estimated from the crop height of alfalfa and other field crops, as is recommended in the Hydrus-1D manual for crop height between 0.10-0.50 m (Simunek et al., 2013).

115

Chapter 4

휌 푐푝 (푒푎 −푒푑) (4-11) 1 훥 (푅푛 − 퐺) 푟푎 퐸푇0 = 퐸푇푟푎푑 + 퐸푇푎푒푟표 = [ 푟 + 푟 ] 휆 훥 + 훾 (1 + 푐 ) 훥 + 훾 (1 + 푐) 푟푎 푟푎

Where ET0 is the evapotranspiration rate (mm/day); ETrad is the radiation (mm/day); ETaero is the aerodynamic (mm/day); λ is the latent heat of vaporization (MJ/kg); Rn is the net radiation

(MJ/m²/day); G is the soil heat flux (MJ/m²/day); ρ is the atmospheric density (kg/m³); cp is the o specific heat of moist air (1.013 kJ/kg/ C); (ea - ed) is the vapor pressure deficit (kPa); ea is the saturation vapor pressure at temperature T (kPa); ed is the actual vapor pressure (kPa); rc is the crop canopy resistance (s/m); ra is the aerodynamic resistance (s/m); Δ is the slope of the vapor pressure curve (kPa/ oC); and γ is the psychrometric constant (kPa/ oC) (Simunek et al., 2013). The actual root water uptake was simulated using Fedde´s root uptake reduction model for grasses, which assumes the water consumption by the crops is related to the water energy status in the soil or substrates (Radcliffe and Simunek, 2010).

푆(ℎ) = 훼(ℎ) 푆푝 (4-12)

Where S(h) is the root water uptake; Sp is the potential root water uptake (1/day); and α(h) is the stress response function of the root water uptake (0 ≤ α ≤1) (Radcliffe and Simunek, 2010). According to Fedde´s model, the maximum potential evapotranspiration rate for grasses, 5 mm/day, is obtained from -25 to -300 hPa, although lower potential evapotranspiration, 1 mm/day, is reached up to -1000 hPa (Simunek et al., 2014). Whereas no evapotranspiration is estimated either when the hydraulic head nears the permanent wilting point, circa -8000 hPa, or is greater than -10 hPa, due to the lack of oxygen in the root depth (Simunek et al., 2014). The relative root distribution was distributed linearly from 1 at the substrate surface (1 cm depth) to zero at the lower profile of the root depth (18 cm depth). Initial pressure head increased linearly from -100 hPa at the substrate surface to zero at the bottom of the substrates. This initial pressure was set because the general field capacity of coarse soils is -100 hPa (Seiler and Gat, 2007; Radcliffe and Simunek, 2010). Observed points were placed at 5 cm, 10 cm, 20 cm, 40 cm, 60 cm, 120 cm, 200 cm, 240 cm and 260 cm depth, where drainage occurs (Figure 4-4).

116

Chapter 4

Figure 4-4: (a) Material distribution; (b) root distribution; (c) initial pressure head; (d) observation points for the forward simulation using Hydrus-1D The upper boundary condition was selected as atmospheric and the lower as seepage face. Atmospheric boundary condition is chosen when the water flux at the substrate air interface is controlled by external conditions, such as precipitation and evaporation (Radcliffe and Simunek, 2010). Seepage face is used on the bottom of non-weighable lysimeters because it considers no water flux when the hydraulic head is negative, however outflow is estimated as soon as the hydraulic head reaches zero (Abichou et al., 2010; Radcliffe and Simunek, 2010). For the Hydrus-1D calibration through the inverse solution, 5 different input data were used to optimize the hydraulic parameters of the substrates (Simunek et al., 2008). Moreover, following the recommendation of Rassam et al. (2003), more than 1 data set to calibrate the model was used (Table 4-3).

Table 4-3: Inverse solution data for the Hydrus 1-D calibration

1 Parameters Seepage Water content ψm (Ɵ) Total Nr. of Nr. Weight Nr. Weight Nr. Weight Observations Inverse 1 104 1 1 1 0 - 105 Inverse 2 4 1 1 1 0 - 5 Inverse 3 - - 19 1 1 10 20 Inverse 4 4 10 19 1 0 - 23 Inverse 5 24 10 19 1 0 - 43 1 ψm 30 hPa The first calibration was performed using 104 weekly observations of the seepage and 1 water content measurement, evaluated on 25.06.2016 from 0.0-0.06 m depth. The second optimization was made using 1 water content measurement and the seasonal accumulated seepage of the substrates (winter and summer 2014, winter and summer 2015). The third calibration considered 19 weekly observed water content measurements (from 25.06.2015 to 29.10.2015), and one retention curve measurement, ψm(Ɵ), weight 10. The fourth included 19 water content measurements and 4 seasonal accumulated seepage measurements, weight 10. The fifth calibration type consisted of 24 monthly observations of the seepage (weight 10), and 19 weekly measurements of water content. 117

Chapter 4

The weight for each measurement represents the importance of the data for calibration (Radcliffe and Simunek, 2010). Different arrangements of fixed and optimized van Genuchten hydraulic parameters were evaluated, however the model converged for all substrates when the volumetric water content at saturation (Ɵs), the saturated hydraulic conductivity (Ks) and the fitting parameters of the water retention curve (n, α) were optimized. The residual water content (Ɵr), was fixed to 0.001 cm³/cm³ and the pore connectivity to 0.5. After running the calibration, the (1) coefficient of determination, R-squared; (2) mass balance error; (3) iteration number and calculation time; (4) root-mean-square error (RMSE); (5) the absolute differences between measured and estimated values of seepage, actual evapotranspiration and water content of the substrates were assessed. The RMSE and the R- squared were also used to evaluate the direct simulation. The coefficient of determination, R-squared, shows how well the optimized model replicates the observed values (Simunek et al., 2013). Mass balance error (%) is the relative error of the water balance fluxes. The number of iterations represents the number of solutions of the global matrix equations to converge the model (Simunek et al., 2013). The root-mean-square error (RMSE) estimates the differences between observed and predicted values (Chai and Draxler, 2014; Schaap et al., 2001). (4-13) 푁 1 푅푀푆퐸 = √ ∑(푂 − 푃 )2 푁 푖 푖 푖=1

Where O and P are the observed and predicted values which refer to the accumulated seepage (mm) or water content (cm³/cm³); N is the number of observations, i.e., 104 seepage values and 24 water content values. The calibrated model that best agreed with the observed values was then used to simulate the water fluxes of the substrates over 27 hydrological years, from the 1st November 1989 to 31st October 2016 using daily weather data from Bad Hersfeld (Deutsche Wetterdienst, 2017a). For this the precipitation height (1-m height), wind speed (12 m height), maximum and minimum air temperature (2-m height), sun hours and air humidity were considered. The sun hours were 2 converted to solar radiation (MJ/m /day) using an ET0 calculator, version 3.2 (FAO, 2014). Missing daily data (61 days from 9862 days) were replaced by the historical monthly means from 1989 to 2016. The Bad Hersfeld weather station, identification number 2171, is located at 50o51´ North, 9o44´ East, 272-m above sea level (Deutsche Wetterdienst, 2017a) and circa 20 km from Heringen, Werra (Deutsche Wetterdienst, 2017b).

118

Chapter 4

4.5.6 Evapotranspiration and drainage of different rates of fine fractions and soil textures Water fluxes were studied using 60 %, 80 % and 100 % fine fractions. For this, the field disturbed samples of the substrates from 0.40 to 0.64 m depth were air dried in the laboratory for circa 10 days. Then the substrates were sieved to obtain fine, < 2 mm of diameter, and coarse fractions, > 2 mm of diameter (Rücknagel et al., 2013). In this sequence, 100 grams of substrate was packed in stainless cylinders according to the different fine fractions based on the oven dried weight. After shaking the cylinders 100 times to settle the substrates, the height covered by the substrates was measured and the substrate´s volume was estimated. Later the samples were saturated for 48 h and oven dried at 105 oC up to a constant weight. Three repetitions were preformed, totaling 36 samples. With the determined bulk densities and the particle size distribution previously estimated with the disturbed samples from the field, the hydraulic parameters of the substrates were found using the Rosetta pedotransfer function (Radcliffe and Simunek, 2010; Schaap et al., 2001), which is implemented in the Hydrus-1D Software, version 4.16.0110 (Simunek et al., 2014; Rassam et al., 2003). The Roseta-based pedotransfer function was also used to determine the hydraulic parameters of different soil textures, such as clay loam, silt loam and sandy loam soils. Further simulations were conducted following the approach suggested by Brakensiek et al. (1986), who state that the saturated hydraulic conductivity decreases according to the relative stone content (stoniness). Whereas, the fitting parameters of the van Genuchten model, i.e. α, m and n, are assumed to be constant (Novák, et al., 2011; Hlaváčiková and Novák, 2014; Beckers et al., 2016).

퐾푠푒 (4-14) 퐾푟 = = 1 − 푅푤 퐾푠

Where Kr is the relative hydraulic conductivity; Kse is the effective saturated hydraulic conductivity of stony soils (cm/d); Ks is the saturated hydraulic conductivity of the fine earth (cm/d);

Rw is the relative stone content in mass units. Moreover, the water content of a stony soil can be adjusted using the expression suggested by Bouwer and Rice (1984):

휃푏 = (1 − 푉푟) 휃푠 (4-15)

Where Ɵb is the bulk volumetric water content of a stony soil; Vr is the volumetric fraction of stones in the medium; and Ɵs is the water content of the soil or substrate. These simulations were performed using the calibrated models from the technogenic substrates installed in the lysimeters.

119

Chapter 4

4.5.7 Statistical Analyses An evaluation of the meteorological data and water balance components was conducted for three hydrological years, 2014, 2015 and 2016. Descriptive statistics were used to summarize the data set from the measurements. The central tendency of the data was determined using mean values, whereas the variability of the mean was determined with standard deviation (SD) and the coefficient of variation (CV) (Crawley, 2014; Field et al., 2012; Couto et al., 2013). Correlation studies were conducted using the Spearman and Pearson correlation coefficients, based on the Shapiro-Wilk normality test (Field, 2013; Field et al., 2012). The correlation and normality tests were performed using RStudio, version 0.99.491 (RStudio Team, 2015; Revelle, 2015).

4.6 Results and discussions In the results section of this paper the hydrological parameters, pH and electrical conductivity of the substrates are presented. The observed and simulated water fluxes are shown by using Hydrus- 1D. Additional results are presented regarding the seepage depths according to different fine fraction rates, soil textures and crop parameters.

4.6.1 Hydraulic parameters of the substrates Figure 4-5 presents an overview of the pH values of substrates 1-4 in 2014 and in 2016. The results indicated that substrate 4 had the highest mean pH level in 2014 and in 2016, 8.7. Whereas the lowest mean pH level was found in substrate 1, 8.4 in 2014 and 2016. A low coefficient of variation among the substrates was found, circa 1.7 % in 2014 and 1.6 % in 2016. The upper boundary showed lower pH levels than the subsurface of the substrates (Figure 4-5). The mean pH at 0.0-0.20 m depth was 8.1 in 2014 and 8.0 in 2016. From 0.20-0.40 m depth, the pH ranged from 8.3 in 2014 to 8.2 in 2016. The lower pH on the surface may be due to the incorporation of organic compost from 0.0 to 0.30 m depth during the installation of the experiment. Additional factors that contribute to the decrease of the pH on the surface are the infiltration of water from precipitation due to its lower pH levels, extraction of cations by the crops and fertilization (Blume et al., 2016). In contrast, an increase trend of the pH from 0.4 to 3.0 m deep was found. This may be due to the leaching of cations from the surface with precipitation water. These results agree with Howard (2017) and Blume et al.’s (2016) findings which showed that technogenic substrates present alkaline pH values due to the presence of coal combustion products, slags, glasses and steel making products. Nevertheless, the pH values recommended for cultivated soils range from 5 and 6.5 (Blume et al., 2016).

120

Chapter 4

12 12

10 10

8 8

6 6

4 4 pH values values 2014 pH values 2016 pH

2 2

0 0 0.0 - 0.2 0.2 - 0.4 1.4 - 1.6 2.0 - 2.2 2.4 - 2.6 2.8 - 3.0 0.0 - 0.2 0.2 - 0.4 1.4 - 1.6 2.0 - 2.2 2.4 - 2.6 2.8 - 3.0 Depth (m) Depth (m) Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1 Substrate 2 Substrate 3 Substrate 4

Figure 4-5: pH values of the substrates 1-4 in 2014 and in 2016 according to different depths The next measurements on electrical conductivity showed that from 0.0 to 2.6 m depth, substrates 1 and 3 presented on average 3.2 mS/cm, substrate 4 3.3 mS/cm and substrate 2 3.5 mS/cm in 2014. In 2016, the electrical conductivity ranged from 2.6 mS/cm in substrate 3 to 2.9 mS/cm in substrate 4. A low coefficient of variation among the substrates was verified for the electrical conductivity in 2014 (3.5 %) and in 2016 (4.3 %). Lower electrical conductivity was measured on the surface and near the surface of the substrates, from 0.0 to 0.4 m depth, which received organic compost, Figure 4-6. The measurements from 2.8 to 3.0 m deep were not included in the aforementioned values, considering that this depth may have been contaminated with the potash tailings residues located on the bottom of the experimental field. According to Allen et al. (1998), perennial ryegrass is moderately tolerant to electrical conductivity, with a threshold of 5.6 mS/cm from the substrate saturation extract. This threshold represents the value in which the crops start to reduce their full potential yield (Allen et al., 1998). In contrast, Niessing (2005) report a limit of 2.25 mS/cm for the electrical conductivity of technogenic substrates to establish a vegetation cover.

121

Chapter 4

7 7 2014 6 2016 6 - -

5 5

4 4

3 3

2 2

1 1 Electrical (mS/cm) conductivityElectrical 0 (mS/cm) conductivityElectrical 0 0.0 - 0.2 0.2 - 0.4 1.4 - 1.6 2.0 - 2.2 2.4 - 2.6 0.0 - 0.2 0.2 - 0.4 1.4 - 1.6 2.0 - 2.2 2.4 - 2.6 Depth (m) Depth (m) Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1 Substrate 2 Substrate 3 Substrate 4

Figure 4-6: Electrical conductivity of substrates 1-4 in 2014 and in 2016 according to different depths The present study also assessed the bulk density of the substrates in 2016 (Figure 4-7), which ranged from 1.17 g/cm³ in substrate 4 to 1.25 in substrate 3. Substrate 1 registered 1.20 g/cm³ and substrate 2, 1.21 g/cm³. The mean bulk density of substrates 1-4 was 1.21 g/cm³. A low variation among the depths (mean 3.9 %) and among the substrates (3.0 %) was verified. The bulk density is one of the main parameters evaluated in evapotranspiration covers because it is associated with root growth, pore space and the water storage capacity of the substrates (Rock et al., 2017; Hauser, 2009). Howard (2017) suggests that the optimal range of bulk density for root growth is 1.3-1.4 g/cm³. Bulk densities higher than 1.5 g/cm³ reduce root growth (Hause, 2009). Additional measurements of the particle density showed values reaching 2.55 g/cm³ in substrate 4 and 2.58 g/cm³ in substrate 1. Almost no variation was found among the depths (0.3 % in substrate 1 and 0.7 % in substrate 4) and among the substrates (0.9 %). The particle densities observed in this investigation are close to the values suggested to mineral soils, 2.6-2.75 g/cm³ (Brady and Weil, 2014). The total porosity estimated from the bulk and the particle density was higher for substrate 4, 54.5 %, and lower for substrate 3, 51 % (Figure 4-7). The mean total porosity of substrates 1-4 was 52.8±1.5 %. A low variation was found for total porosity among the depths, 3.6 %, and among the substrates, 2.7 %. Nevertheless, the mean total pore volume by saturation for substrates 1-4, 48.6 ± 3.0 %, was lower than the estimated values. The differences between the estimated and measured total porosity can be related to the entrapped air within the samples during saturation.

122

Chapter 4

2.0 60 1.8 50 1.6 1.4 40 1.2 1.0 30 0.8 20 0.6 Total porosity porosity (%) Total Bulk density Bulk (g/cm³) 0.4 10 0.2 0.0 0 0.0 - 0.04 0.20 - 0.24 0.40 - 0.44 0.60 - 0.64 0.0 - 0.04 0.20 - 0.24 0.40 - 0.44 0.60 - 0.64 Depth (m) Depth (m) Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1 Substrate 2 Substrate 3 Substrate 4

Figure 4-7: Bulk density and total porosity of the substrates 1-4 in 2016 according to different depths In the experimental study of the water retention curve a similar desaturation process among the substrates was observed. The water retention curve of substrate 1 from 0.0 to 0.64 m is shown in Figure 4-8, substrates 2-4 presented a similar trend (Figure 4-24 - Figure 4-26, supplementary materials). The lowest water volume released by gravity was found in substrate 3, circa 13.2 % whereas the maximum was registered in substrate 1, 16.4 %. The available plant water ranged from 18.4 % in substrate 4 to 22.2 % in substrate 1. In addition, the unavailable moisture varied between 11.7 % (substrate 1) and 14.1 % (substrate 4), Table 4-11, supplementary materials. Further analyses on the pore size distribution demonstrated that, on average, the large coarse pores totaled 14.9 % of the total substrates volume, tight coarse pores 6.7 %; medium pores 11.0 %, and fine pores 14.1 %, Table 4-12, supplementary materials. Midsize pores are desirable in evapotranspiration covers because these pores can hold water against gravity and plants can easily remove the water from them (Hauser, 2009).

123

Chapter 4

0.6 ) 3 S1 (0.0 - 0.04 m) /cm 3 0.5 S1 (0.20 - 0.24 m) S1 (0.40 - 0.44 m) S1 (cm S1

- 0.4 S1 (0.60 - 0.64 m)

0.3 S1 (0.0 - 0.04 m) S1 (0.20 - 0.24 m) 0.2 S1 (0.40 - 0.44 m) S1 (0.60 - 0.64 m) 0.1

Volumetric water Volumetric content 0.0 1 10 100 1000 10000 100000 1000000 10000000 Matric potential (hPa, log. scale)

Figure 4-8: Water retention curve from substrate 1 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6) The saturated hydraulic conductivity measured with the falling head method presented on average 748.3 cm/d at 0.0-0.04 m depth; 914.6 cm/d at 0.20-0.24 m depth, 453.3 cm/d at 0.40-0.44 m depth and 305.1 cm/d at 0.60-0.64 m depth for substrates 1-4 (Figure 4-9). The low trend of saturated hydraulic conductivity according to the depths can be associated with an increase in bulk density. In addition, the observed high variation among the replicates (Figure 4-9) can be attributed to the coarse pores of the samples due to the different sizes and shapes of the particles (Howard, 2017; Brady and Weil, 2014). According to the Manual of Soil Mapping (AG Boden, 2005), the measured saturated hydraulic conductivities in this study are considered extremely high. Niessing (2005) mentions that high levels of hydraulic conductivity may minimize the water erosion of evapotranspiration covers on potash tailings piles. Though, high hydraulic conductivity can increase the leaching of chemicals (van Genuchten, 1992).

124

Chapter 4

1800 1200

1600 (a) (b) 1000 1400

1200 800 1000

800 600

600 (cm/d) 400 400

200 200

0 Saturated hydraulic (cm/d) conductivitySaturated hydraulic

0.0 - 0.04 0.20 - 0.24 0.40 - 0.44 0.60 - 0.64 conductivityhydraulic Meansaturated 0 0.0 - 0.04 0.20 - 0.24 0.40 - 0.44 0.60 - 0.64 Depth (m) Substrate 1 Substrate 2 Substrate 3 Substrate 4 Depth (m)

Figure 4-9: (a) Mean saturated hydraulic conductivity for each substrate and depth (n=6); (b) mean saturated hydraulic conductivity for the substrates 1-4 at different depths (n=24) With the Kostiakov infiltration model of the substrate 1 (I = 4.102*T0.7418, R²: 0.9866), substrate 2 (I = 5.1875*T0.7005, R²: 0.9819), and substrate 4 (I = 6.6529*T0.7681, R² = 0.998) saturated hydraulic conductivity of 30.7 mm/h (73.7 cm/d) for substrate 1, 28.7 mm/h (68.9 cm/d) for substrate 2, and 53.5 mm/h (128.4 cm/d) for substrate 4 were estimated. The accumulated infiltration recorded for substrate 3 (I = 4.5899*T0.5723, R² = 0.9843, 12.7 mm/h, 30.5 cm/d) was not considered. As the saturated hydraulic conductivity measurement was done in the outer part of the experimental site, a very high coefficient of variation was verified among the substrates, 36.7% (mean: 37.6 mm/h). These measured saturated hydraulic conductivity values are lower than the ones recorded in the laboratory with the falling head method (Figure 4-9). The differences between field and laboratory measurements for the saturated hydraulic conductivity can be attributed to compaction and entrapped air in the substrates during the double ring infiltration tests (Lal and Shukla, 2004).

125

Chapter 4

200 200

0.7005 180 y = 4.102x0.7418 180 y = 5.1875x R² = 0.9819 160 R² = 0.9866 160 S2 (mm) S2

S1 (mm) S1 140

140 - - 120 120 100 100 80 80 60 60 40 40 Cumulative infiltration Cumulative infiltration Cumulative infiltration Cumulative infiltration 20 20 0 0 0 20 40 60 0 20 40 60 Cumulative time (min) Cumulative time (min)

200 y = 6.6529x0.7681 180 R² = 0.998 160

S4 (mm) S4 140 - 120 100 80 60 40

Cumulative infiltration Cumulative infiltration 20 0 0 20 40 60 Cumulative time (min)

Figure 4-10: Accumulated infiltration of substrates 1, 2 and 4 in 2014 The analysis of the particle size distribution from the fine fraction of the substrates (< 2 mm) revealed that the substrates are classified as sandy loam (Blume et al., 2016), with on average 52 % sand-size particles (0.063-2 mm), 43 % silt-size particles (2-63 µm) and 5 % clay-size particles (< 2 µm), Table 4-14, supplementary materials. In addition, the substrates showed a coarse fraction of 42 % (Ø > 2 mm), Figure 4-11a.

126

Chapter 4

100 100 (a) (b) 90 90 80 80 70 70 60 60

50

% Mass

Figure 4-11: (a) Cumulative grain size distribution curves of fine and coarse particles from substrates 1-4; (b) Cumulative grain size distribution curves of fine particles from substrates 1-4. Mean values from 0.0 to 3.0 m depth

From the cumulative grain size distribution curves (Figure 4-11) the coefficient of uniformity from the substrates were estimated (Horton et al., 2016). The coefficient of uniformity of the fine and coarse particles, ranged from 387.5 in substrate 1 to 416.7 in substrate 2 and 4, totaling a mean of 399 and a coefficient of variation of 5.3 %, which is considered low (Table 4-4). Higher variation was found when studying the fine fraction (< 2 mm diameter). For the fine fraction, the coefficient of uniformity ranged from 41.4 in substrate 3 to 75.0 in substrate 4, totaling a mean of 57.4 and a coefficient of variation of 27.9 %, which is considered high. According to Selker et al. (1999), the uniformity coefficients of well sorted materials is around 2 and poorly sorted around 10. For Horton et al. (2016) the coefficient of uniformity indicates the behavior of the materials when receiving water. More water filling the voids for non-uniform distributed grains is expected (Horton et al., 2016). Large numerator values in the uniformity coefficient indicate that the soil may settle under pressure and that the hydraulic conductivity may change due to the filling of voids (Horton et al., 2016). Uniformity coefficients are important for construction and compaction purposes (Horton et al., 2016). Soils with a uniformity coefficient lower than 5 are considered stable and rigid soils, whereas values larger than 15 indicate materials that settle under load (Horton et al., 2016).

127

Chapter 4

Table 4-4: Median equivalent diameter, d50, and uniformity coefficients of substrates 1-4 for the entire substrates (diameter < 12 mm) and for the fines (diameter < 2 mm)

Fine and coarse particles Fine particles Substrates d50 U d50 U μm - μm - Substrate 1 2000.0 387.5 120 66.7 Substrate 2 950.0 416.7 50 46.7 Substrate 3 600.0 375.0 70 41.4 Substrate 4 1000.0 416.7 90 75.0 Mean 1137.5 399.0 82.5 57.4 Standard deviation 601.9 21.1 29.9 16.0 Coefficient of variation 52.9 5.3 36.2 27.9

The median equivalent diameter, d50, which represents the diameter value that separate the grain size distribution in two shares (Horton et al., 2016) of the substrates 1-4 is shown in Table 4-4. On average the fine particles had 50 % of smaller grains than 82.5 μm and the mixed fine and coarse particles showed a median diameter of 1137.5 μm. Further studies on the color of the substrates revealed that the substrates are dark grey (10YR3/1) for water content ranging from 20-30 %, or gray (10YR5/1) for oven dried substrates. Dark soils absorb more heat and increase evaporation because they have more available energy (Geiger et al., 2009). The evaluation of the weekly water content of the substrates in Heringen from 25.06.2015 to 27.10.2016 showed that the volumetric water content was generally lower than the saturation rate measured in the laboratory (48.6 %, Table 4-11, suplementary materials). Moreover, summer weeks registered a lower water content with a mean of 16.9 %, than the winter weeks, mean 30.6 %, Figure 4-12.

128

Chapter 4

60 90 80 50 70 40 60 50 30 40 20 30 20 Precipitation (mm/week) Precipitation

Water content (cm³/cm³; (cm³/cm³; %) Water content 10 10 0 0 7-Jul-2016 8-Oct-2015 8-Sep-2016 16-Jul-2015 3-Mar-2016 28-Jul-2016 6-Aug-2015 21-Jan-2016 5-May-2016 25-Jun-2015 16-Jun-2016 29-Oct-2015 20-Oct-2016 17-Sep-2015 11-Feb-2016 29-Sep-2016 14-Apr-2016 10-Dec-2015 31-Dec-2015 24-Mar-2016 27-Aug-2015 19-Nov-2015 18-Aug-2016 13-May-2015 25-May-2016 Time (date) Precipitation Substrate 1 Substrate 2 Substrate 3 Substrate 4

Figure 4-12: Weekly water content of substrates 1-4 from 25.06.2015 to 27.10.2016

4.6.2 Observed water fluxes of the substrates The annual ground level precipitation is shown in Table 4-1. The measured seepage and the estimated actual evapotranspiration of substrates 1-4 are shown in Figure 4-13.

600 600 500 500 400 400 300 300 523 527 200 449 200 437 370 346 Substrate 1 Substrate (mm) Substrate 2 (mm) Substrate - 100 - 100 0 0

-100 174 -100 197 265 235 261 247 -200 -200 Water fluxes Water fluxes Water fluxes Water fluxes -300 -300 -400 -400 2014 2015 2016 2014 2015 2016

Time (water years) Time (water years)

129

Chapter 4

600 600 Actual evapotranspiration 500 500 Drainage 400 400

300 300 513 464 505 462 200 200 347 344 Substrate 4 Substrate (mm) 100 -

Substrate 3 Substrate (mm) 100 - 0 0

-100 197 220 -100 200 222 275 283

-200 Water fluxes -200 Water fluxes Water fluxes -300 -300

-400 -400 2014 2015 2016 2014 2015 2016 Time (water years) Time (water years)

Figure 4-13: Observed water fluxes of substrates 1-4 during 2014, 2015 and 2016 (± standard deviation) The highest seepage was registered in 2014, ranging from 261.2 mm in substrate 2 to 282.9 mm in substrate 4 (mean substrate 1 - substrate 4 271.2 ± 9.8 mm). The lowest seepage was found in 2015, which varied from 173.9 mm in substrate 1 to 200.2 mm in substrate 4 (mean substrate 1 - substrate 4 192.1 ± 12.2 mm). In 2016, 235.0 mm of seepage was registered in substrate 1, 247.1 mm in substrate 2, 220.2 mm in substrate 3, and 222 mm in substrate 4 (mean substrate 1 - substrate 4 231.1 ± 12.6 mm). A low coefficient of variation was observed for the drainage among the substrates, which was 3.6 % in 2014, 6.4 % in 2015 and 5.5 % in 2016. The ratio between the drainage and ground-level precipitation (D/P), was on average 34.4 % in 2014, 35.3 % in 2015 and 33.8 % in 2016. Following the trend of the seepage, the actual evapotranspiration was the highest in 2014, which ranged from 505.3 mm in substrate 4 to 527.1 mm in substrate 2 (mean substrate 1 - substrate 4 517.1 ± 1.9 mm). The lowest was in 2015 when substrate 1 registered 369.9 mm, substrate 2 346.4 mm, substrate 3 346.9 mm, and substrate 4 343.6 mm (mean 351.7 ± 3.5 mm). In 2016, the actual evapotranspiration was 448.6 mm in substrate 1, 436.9 mm in substrate 2, 463.7 mm in substrate 3 and 462.1 mm in substrate 4 (mean substrate 1 - substrate 4 452.8 ± 2.8 mm). The actual evapotranspiration registered a low variation among the substrates in 2014 (1.9 %), 2015 (3.5 %) and 2016 (2.8 %). The ratio between actual evapotranspiration and precipitation, was on average 65.6 % in 2014, 64.7 % in 2015 and 66.2 % in 2016. As discussed already by Bilibio et al. (2017), the low variation of water fluxes indicates that although the substrates were made of different proportions of coal combustion residues and household incineration slags, these proportions were not enough to lead to significant changes in the seepage or actual evapotranspiration rates. 130

Chapter 4

4.6.3 Forward simulation, calibration and validation of the Hydrus-1D model The direct simulation of water fluxes using Hydrus-1D was performed using hydraulic properties from the substrates and weather data from 2014-2015. The results are presented in Table 4-15, Table 4-20 - Table 4-29, supplementary materials. The simulations presented a high association with the observed seepage and moderate to water content values (Table 4-15, supplementary materials). However, the absolute differences between measured and predicted seepage were found to be high. For instance, substrate 1 registered 23.0 mm less than the total observed seepage, substrate 2 -93.7 mm; substrate 3 -94.3 mm; substrate 4 -90.5 mm, comprising a mean variation of -16 %. The higher RMSE was found in substrate 3, 52.2 mm, and the lowest for substrate 1, 34.8 mm. Considering these results, the Hydrus-1D model was calibrated according to different inverse solution data (Table 4-3). The initial estimates and optimized hydraulic parameters of the substrates 1-4 according to the different calibrations are shown in Table 4-16 - Table 4-19, supplementary materials. A high association between calibrated and measured values (R² > 0.9) was observed for all calibration methods, except when using the water content and retention curve measurements (0.8 ≤ R² ≤ 0.92). Further analyses revealed the number of iterations were within the initial configuration of the model, up to 20 iterations. The calculation time of the inverse solution ranged from 19 to 53.9 seconds, which is considered low. In addition, the water mass balance errors were lower than 1 %, which is the upper limit for water balance assessments (Radcliffe and Simunek, 2010). The estimation of the root-mean-square error (RMSE) registered the highest value for the calibration using water content and retention curve values (inverse solution 3), comprising 75.9 mm for substrate 1; 163.1 mm for substrate 2; 51.5 mm for substrate 3; and 35.1 mm for substrate 4. When the seepage observations were used in the inverse simulation, the RMSE was similar among the models. However, the inverse simulation using 104 measurement of the seepage and 1 reading of the water content presented the lowest RMSE for the water content, consisting of 0.065 cm³/cm³ in substrate 1; 0.058 cm³/cm³ in substrate 2; 0.079 cm³/cm³ in substrate 3; and 0.091 cm³/cm³ in substrate 4. The absolute differences between predicted and observed values of the accumulated seepage and actual evapotranspiration were also studied. The highest differences were observed when using water content and retention curve measurements (calibration type number 3). With this calibrated model, substrate 1 registered 27 % less drainage than the total observed volume. Likewise, substrate 2 estimated 60 %; substrate 3 15 %; and substrate 4 12 % less seepage. Lower differences between the predicted and observed seepages were estimated when the inverse simulation was performed using drainage measurements. For instance, when using 104 drainage measurements in the Hydrus inverse solution, the differences between observed and predicted values ranged from -1.3% in 131

Chapter 4 substrate 1 to -5.1% in substrate 4 (mean -2.9 %, -13.6 mm). Nevertheless, the inverse simulations using 104 drainage measurements predicted a higher total actual evapotranspiration, circa 7.3 % (63.3 mm) more than the observed measurements (Eq. 4-2). These differences can be related to salt stress (Simunek et al., 2013), nutrient availability (Blume et al., 2016) or any change in the grass cover in the field (Allen et al., 1998), i.e., natural integration of native crops (Papke and Schmeisky, 2013). While assessing the hydraulic parameters of the substrates, the inverse simulation increased the values of the fitting parameters from van Genuchten (1980) model, alpha and n, which made the water retention curves move to the left (Figure 4-14). These differences between observed and optimized fitting parameters of the retention curve may be related with the use of fine fractions on the edges of the cylinders in the field to improve the contact with the suction ceramic plates in the laboratory. An increase trend of the fitting parameters of the water retention curve is expected when using coarser materials (Rassam et al., 2003; Schaap et al., 2001). Moreover, the differences in size of the flow domain (cylinder samples and 2.6 m depth lysimeters) and the wetting and drying processes in the field – hysteresis (Hopmans, 2010), can also contribute to differences between measured and optimized fitting parameters of the retention curve (Kodešová et al., 2014). For the saturated hydraulic conductivity, the inverse simulations provided lower values than the ones measured in the laboratory. For example, the mean saturated hydraulic conductivity of the substrate 1 from 0.0 to 0.64 m depth was 687.4 cm/d, whereas the optimized parameter was 182.14 cm/d. This may be due to errors in the laboratory measurements, such as water flow in the internal walls of the cylinders and sample size (Shukla, 2014). The optimized saturated hydraulic conductivities were however higher than the values measured with the double ring infiltrometer. Taking into account the differences between observed and predicted seepage, actual evapotranspiration and water content, the inverse simulation conducted with 104 observations of seepage and 1 measurement of water content was used to carry out further studies. Figure 4-15 shows the observed versus predicted accumulated seepage values from substrates 1-4 using the mentioned inverse simulation. Differences between predicted and observed seepage can be attributed to macropores, cracks and lateral flow on the wall of the lysimeters (Li et al., 2014) which are not considered in the Richards equation (Hendriks, 2010; van Genuchten, 1992). Figure 4-16 shows the accumulated and daily potential and actual root water uptake from substrate 1 predicted using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D over two hydrological years. Moreover, Figure 4-17 highlights the predicted daily water content in substrate 1 at different observation points. Substrates 2-4 presented similar values. And Figure 4-18 illustrates the water storage change of substrates 1-4 in the entire flow domain of the lysimeters from 0.0 to 2.6 m depth.

132

Chapter 4

0.6 0.6 ) 3 ) 3 /cm

3 0.5 0.5 /cm 3 S1(cm

0.4 S2(cm 0.4 - -

0.3 0.3

0.2 0.2

0.1 0.1 Volumetric content Volumetric water Volumetric content Volumetric water 0.0 0.0 1 100 10000 1000000 1 100 10000 1000000 Matric potential (hPa, log. scale) Matric potential (hPa, log. scale)

0.6 0.6 Observed model (0.0 - 0.64 m) ) ) 3 3 Optimized model (0.0 - 2.6 m)

/cm 0.5 0.5 /cm 3 3

S3(cm 0.4 0.4 S4(cm - -

0.3 0.3

0.2 0.2

0.1 0.1 Volumetric content Volumetric water Volumetric content Volumetric water 0.0 0.0 1 100 10000 1000000 1 100 10000 1000000 Matric potential (hPa, log. scale) Matric potential (hPa, log. scale) Figure 4-14: Observed and optimized water retention curve from substrates 1-4. The solid line is the observed model from 0.0 to 0.64 m depth. The dashed line is the optimized model from 0.0 to 2.60 m depth using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D over two hydrological years

133

Chapter 4

0 0 -50 -50 -100 -100 -150 -150 -200 -200 -250 -250 -300 -300 -350 -350 -400 -400 -450 -450 -500 -500 Accumulated seepage substrate (mm) substrate 1 seepage Accumulated (mm) substrate 2 seepage Accumulated 3-Apr-14 4-Dec-14 10-Jul-14 7-Nov-13 6-Aug-15 10-Jul-14 22-Jan-15 18-Jun-15 16-Oct-14 22-Jan-15 13-Feb-14 24-Sep-15 18-Jun-15 30-Apr-15 16-Oct-14 27-Dec-13 13-Feb-14 24-Sep-15 12-Mar-15 03-Apr-14 30-Apr-15 27-Dec-13 04-Dec-14 28-Aug-14 12-Mar-15 22-May-14 07-Nov-13 28-Aug-14 06-Aug-15 22-May-14 Time (date) Time (date)

0 0 -50 -50 Observed values -100 -100 Predicted values -150 -150 -200 -200 -250 -250 -300 -300 -350 -350 -400 -400 -450 -450 -500 -500 Accumulated seepage substrate (mm) substrate 3 seepage Accumulated (mm) substrate 4 seepage Accumulated 10-Jul-14 22-Jan-15 18-Jun-15 3-Apr-14 16-Oct-14 4-Dec-14 13-Feb-14 24-Sep-15 10-Jul-14 03-Apr-14 30-Apr-15 7-Nov-13 6-Aug-15 27-Dec-13 04-Dec-14 12-Mar-15 22-Jan-15 07-Nov-13 28-Aug-14 06-Aug-15 18-Jun-15 16-Oct-14 22-May-14 13-Feb-14 24-Sep-15 30-Apr-15 27-Dec-13 12-Mar-15 28-Aug-14 22-May-14 Time (date) Time (date) Figure 4-15: Observed versus predicted accumulated drainage of substrates 1-4 using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D during two hydrological years

134

Chapter 4

2000 12 2014 2015 2014 2015 10 1600

8 1200 6 800 4

400 2 Potential root uptake (mm) root Potential water 0 0 Potential root uptake (mm/day) root Potential water 1-Jul-14 1-Jul-15 1-Jul-14 1-Jul-15 1-Jan-14 1-Jan-15 1-Jan-14 1-Jan-15 1-Sep-14 1-Sep-15 1-Sep-14 1-Sep-15 1-Mar-14 1-Mar-15 1-Mar-14 1-Mar-15 1-Nov-13 1-Nov-14 1-Nov-13 1-Nov-14 1-May-14 1-May-15 1-May-14 1-May-15 Time (date) Time (date)

2000 12 2014 2015 2014 2015 10 1600

8 1200 6 800 4

400 2 Actual root water uptake (mm) uptake root water Actual

0 uptake (mm/day) root Actual water 0 1-Jul-14 1-Jul-15 1-Jul-14 1-Jul-15 1-Jan-14 1-Jan-15 1-Jan-14 1-Jan-15 1-Sep-14 1-Sep-15 1-Sep-14 1-Sep-15 1-Mar-14 1-Mar-15 1-Nov-13 1-Nov-14 1-Mar-14 1-Mar-15 1-May-14 1-May-15 1-Nov-13 1-Nov-14 1-May-14 1-May-15 Time (date) Time (date) Figure 4-16: Predicted potential and actual root water uptake of substrate 1 using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D over two hydrological years

135

Chapter 4

0.6 60 2014 2015 0.5 50

0.4 40

0.3 30

0.2 20

0.1 10 Precipitation (mm/day) Precipitation

0 0 Volumetric water Volumetric content(cm³/cm³) 1-Jul-2014 1-Jul-2015 1-Jan-2014 1-Jan-2015 1-Jun-2014 1-Jun-2015 1-Oct-2014 1-Oct-2015 1-Feb-2014 1-Sep-2014 1-Feb-2015 1-Sep-2015 1-Apr-2014 1-Apr-2015 1-Dec-2013 1-Dec-2014 1-Mar-2014 1-Mar-2015 1-Nov-2013 1-Aug-2014 1-Nov-2014 1-Aug-2015 1-May-2014 1-May-2015 Time (date) Precipitation 0.05 m 0.10 m 0.20 m 0.40 m 0.60 m 1.20 m 2.00 m 2.40 m 2.60 m Observed (0 - 0.06 m) Figure 4-17: Predicted water content levels in different observation points of substrate 1 using 104 observations of seepage and 1 measurement of water content in the inverse solution of Hydrus-1D during two hydrological years

1200 2014 2015 1000

800

600

400

Water storage Water (mm/day) storage 200

0 1-Jul-14 1-Jul-15 1-Jan-14 1-Jan-15 1-Jun-14 1-Jun-15 1-Oct-14 1-Oct-15 1-Feb-14 1-Sep-14 1-Feb-15 1-Sep-15 1-Apr-14 1-Apr-15 1-Dec-13 1-Dec-14 1-Mar-14 1-Mar-15 1-Nov-13 1-Aug-14 1-Nov-14 1-Aug-15 1-May-14 1-May-15 Time (date) Substrate 1 Substrate 2 Substrate 3 Substrate 4 Figure 4-18: Predicted water storage of the substrates 1-4 in the entire flow domain of the lysimeters, from 0.0 to 2.6 m depth

136

Chapter 4

The accumulated and daily actual root water uptake (mean S1-S4 932.4 mm; 1.3 mm/d) of the substrates were generally lower than the potential root water uptake (mean S1-S4 1766.7 mm; 2.4 mm/d). This is because the potential root water uptake considers no limitation in water and nutrients (Allen et al., 1998). Additionally, higher daily values of potential root water uptake and actual root water uptake were observed in summer months, which registered higher radiation and temperature levels (Figure 4-3). The mean potential root water uptake in the summer was 3.14 mm/d, and the actual root water uptake, 1.58 mm/d. Whereas in winter the mean potential root water uptake was 1.69 mm/d, and the actual root water uptake, 0.95 mm/d for the substrates 1-4. The daily water content levels in different observation points (Figure 4-17) shows that the water content oscillates up to 0.6 m depth. In deeper layers the water content is more constant. The substrates’ surface is more exposed to atmospheric conditions, such as precipitation, wind and temperature (Kodešová et al., 2014). In addition, the root water uptake takes place near the surface, which contributes to the variability of the water status in this region (Radcliffe and Simunek, 2010; Lal and Shukla, 2004). From the daily water storage of substrates 1-4, Figure 4-18, the water storage on the substrates increases in winter and decreases in summer, following the pattern of the root water uptake. The larger increase of the water storage in July 2014 may be due to the accumulated rain in this month. Ground-level gauges registered 237.6 mm and the Thies weather station, 221.6 mm, in July 2014. During the respective period, the stored water increased 113.2 mm in substrate 1, 110.7 mm in substrate 2, 115.2 mm in substrate 3 and 117.0 mm in substrate 4. The storage capacity of the substrates, i.e. hydraulic conductivity near to 0.01 cm/d (Simunek et al., 2014), for 2.6 m deep of substrates is: 721.8 mm in substrate 1 (277.6 mm/m); 647.73 mm in substrate 2 (249.1 mm/m); 589.1 mm in substrate 3 (226.6 mm/m); and 555.7 mm in substrate 4 (213.7 mm/m) according to the calibrated model. The matric potential of the substrates at field capacity, is 103.7 hPa for substrate 1; 101.3 hPa for substrate 2; 80.4 hPa for substrate 3; and 71.7 hPa for substrate 4. These values are within the range of the field capacity of mineral soils, pF 1.8 to pF 2.5 (Blume et al., 2016).

4.6.4 Validation of the calibrated Hydrus-1D model and predictions Due to the lack of precipitation measurements from the Thies weather station in 2016, the daily weather data available from Bad Hersfeld was used to validate the calibrated model. 27 water years, from 1990 to 2016 were used. For this simulation, a constant root depth of 18 cm and a crop height of 30 cm were used. Figure 4-19 shows the accumulated precipitation, drainage, potential and actual root water uptake for Bad Hersfeld using the optimized hydraulic parameters from substrate 1.

137

Chapter 4

2200

1800

1400

1000

600

200

-200 Cumulative water fluxes fluxes Cumulative (cm) water -600 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Time (water years) Precipitation Drainage Potential root water uptake Actual root water uptake

Figure 4-19: Accumulated water fluxes for 27 water-years for Bad Hersfeld using optimized hydraulic parameters from substrate 1 From Figure 4-19, an accumulated precipitation of 1809.9 cm (mean 67 cm/yr) was observed; potential root water uptake of 2004.3 cm (mean 74.2 cm/yr); actual root water uptake of 1360.8 cm (mean 50.4 cm/yr); and an accumulated seepage of 429.4 cm (mean 15.9 cm/yr). Moreover, the total evaporation was 14.9 cm over 27 years (mean 0.6 cm/yr). Considering the ratio between seepage and precipitation (D/P), a seepage rate of 23.7 % was estimated using the hydraulic parameters of substrate 1, 23.7 % for substrate 2; 25.3 % for substrate 3; and 26.1 % for substrate 4. These results agree with Hermsmeyer (2001), who found a seepage rate of 24.4 % from a technogenic substrate made of aluminium recycling by-products (70 %) and flue gas desulfurization by-product (30 %), over a potash tailings pile located near Hannover, Germany. However, if the crop height is decreased from 30 to 20 cm, a mean seepage rate of 27.3 % is obtained. Also, if the root depth is increased to 28 cm, and the crop maintains a height of 30 cm, the calibrated model predicts a mean seepage rate of 22.2 %, which is slightly lower than the simulations using a root depth of 18 cm (mean S1-S4 24.7 %). This shows that the water fluxes of the evapotranspiration covers are affected by any change in the crop status. The crop height is used to estimate leaf area index and aerodynamic resistances in the Penmann-Monteith equation (Simunek et al., 2013). Whereas the root depth represents the substrates area explored by the roots and from where the water is transported to the atmosphere (Radcliffe and Simunek, 2010). The seepage rates according to different root depths and crop height for 27 water-years for Bad Hersfeld using optimized hydraulic parameters from substrates 1 are shown in Figure 4-20.

138

Chapter 4

35 35

30 30 26.6 26.3 25 23.1 25 23.7 21.6 20.7 19.8 20 18.8 20 18.2 17.1 16.7 15.5 15.3 14.0 14.1 15 12.5 15 12.9 11.2 9.9 Seepage rate (%) Seepage rate Seepage rate (%) rate Seepage 10 10

5 5 Seepage rate Seepage rate 0 0 10 20 30 40 50 60 70 80 90 100 20 30 40 50 60 70 80 90 100

Root depth (cm) Crop height (cm)

Figure 4-20: Seepage rates according to different root depths and crop height for 27 water-years for Bad Hersfeld using optimized hydraulic parameters from substrates 1 By looking closer at the water years, one observes a variation of circa 38.9 % for the seepage and 20.5 % for the actual evapotranspiration (Table 4-31, supplementary materials). This variation is higher than the ones observed during the three water years in Heringen, 17.1 % for drainage and 18.9 % for the actual evapotranspiration among the hydrological years from 2014 to 2016. However, this may be expected due to the variation in precipitation, solar radiation, temperature and cloudiness during the growing season of the perennial grasses.

4.6.5 Water fluxes using different rates of fine fractions and soil textures The average values of bulk density from substrates 1-4 using different fine fraction rates are shown in Figure 4-21. The bulk density ranged from 1.13 g/cm³ in the substrates free of fine particles to 1.32 g/cm³ in the substrates with 40 % of fine particles. An increase trend of bulk density was verified from 100 to 40 % of fine particles. These results agree with Fiès et al. (2002), who reported an increase trend for the bulk density in clay and clay- silt soils when using up to 50 % mass weight of coarse glass content. Horton et al. (2016) state that the densest packing is found in poorly sorted materials.

139

Chapter 4

1.40 y = -0.0253x2 + 0.186x + 0.9736 1.35 R² = 0.9746

1.30

1.25

1.20

1.15 1.32 1.31 1.25 1.25

Dry (g/cm³) Dry bulk density 1.10 1.19 1.13 1.05

1.00 0 20 40 60 80 100 Fine fraction (%)

Figure 4-21: Dry bulk density of the substrates 1-4 according to different rates of fine fraction (± standard deviation) With the bulk and particle density of the substrates the total porosity was estimated, which ranged from 48.5 % to 55.9 % (Figure 4-22). The increase of total porosity when using higher rates of coarse fraction may be due to non-filled air space between coarse and fine particles, which leads to large pores and channels (Lal and Shukla, 2004; Fiès et al., 2002).

60 y = 0.9955x2 - 7.2833x + 61.832 55 R² = 0.9667

50

45

40 55.9 53.5 50.7 51.0 35 48.5 48.9 Total porosity porosity (%) Total 30

25

20 0 20 40 60 80 100 Fine fraction (%)

Figure 4-22: Total porosity of the substrates 1-4 according to different rates of fine fractions (± standard deviation) The water fluxes were simulated using 60, 80 and 100 % of fine particles. The hydraulic parameters and the respective seepages simulated using Hydrus-1D are presented in Table 4-5. Table 4-5 also show the simulations using three classes of soil textures, i.e., clay loam, silt loam, and sandy loam.

140

Chapter 4

Table 4-5: Mean hydraulic parameters and accumulated seepage using 60 %, 80 % and 100 % fine fraction of substrates 1-4 and soil textures from 2014 to 2015 in Heringen

Fine rate Ɵr Ɵs α n Ks Volume¹ Total seepage Differences cm³/cm³ cm³/cm³ 1/cm cm/d mm mm % mm % 100 % 0.0373 0.4066 0.0115 1.5014 101.2 459.6 259.6 20.2 80 % 0.0362 0.3934 0.0126 1.4923 81.3 458.3 291.0 22.6 31.5 12.1 60 % 0.0352 0.3820 0.0141 1.4815 65.7 459.5 332.7 25.9 73.1 28.2 Clay loam 0.095 0.41 0.019 1.31 6.24 768.3 391.9 30.5 Silt loam 0.067 0.45 0.02 1.41 10.8 708.5 421.8 32.8 29.9 7.6 Sandy loam 0.065 0.41 0.075 1.89 106.1 362.7 520.4 40.4 128.5 32.8 1 Soil water storage at field capacity in 2.6 m deep Regarding the retention curve parameters, a decrease trend for the water content at saturation was verified as well as for the saturated hydraulic conductivity according to the decrease of the fine fraction rate (Table 4-5). The water retention curve of the substrates with 60, 80 and 100 % fine particles is shown graphically in Figure 4-23a. For the seepage, an accumulated predicted seepage rate of 332.7 mm was observed using 60 % of fine particles; 291.0 mm using 80 % of fine particles; and 259.6 mm using 100 % fine particles in Heringen from 2014 to 2015.The use of 60 % of particles lower than 2 mm presented the highest difference with 100 % of fine earth, 73.1 mm or a relative variation of 28.2 %. A similar pattern was found for the simulated water fluxes using different soil textures (Table 4-5). Clay loam registered the lowest total seepage, 391.9 mm, whereas the highest was found using sandy loam, consisting of 520.4 mm. This trial shows that on average the packed samples presented different water fluxes from the measured and calibrated models. For instance, the seepage registered in Heringen from 2014 to 2015 was 439.1 mm for substrate 1, 458.6 mm for substrate 2, 472.3 mm for substrate 3 and 483.1 mm for substrate 4, whereas the simulation using 60 % fine particles in the packed sample was 332.7 mm (mean substrates 1-4). These differences can be due to the packing method which affected the dry bulk density (the samples were shaken 100 times to settle the substrates) and lead to different water retention parameters. More precise hydraulic parameters from the Rosetta pedotransfer function may be obtained by increasing the number of predictors, i.e., water content at suction of 33 and 1500 kPa (Schaap et al., 2001). However, previous test in the laboratory (data not shown) presented even higher differences than the current evaluation using dry bulk density and texture. When applying the approach of relative stone content for the water retention curve parameters (Brakensiek et al., 1986; Novák, et al., 2011; Hlaváčiková and Novák, 2014; Beckers et al., 2016) of the calibrated model from substrate 1, a decreasing trend for the seepage was also found when increasing the rate of fine fraction from 40 % to 80 % (Table 4-6), in Bad Hersfeld over 27 years of

141

Chapter 4

historical weather data. Further simulations using 100 % fine fractions were not performed because the water content at saturation when using 80 % of fine earth was close to the upper limit of air space in mineral soils or substrates, which is circa 0.6 cm³/cm³ (Blume et al., 2016), Figure 4-23b.

Table 4-6: Accumulated seepage using the relative stone content approach for substrate 1 from 1990 to 2016 in Bad Hersfeld

1 Fine earth Ɵr Ɵs α n Ks Volume Total seepage Differences cm³/cm³ cm³/cm³ 1/cm cm/d cm cm % mm % 80 % 0.0 0.600 0.0893 1.2597 218.6 85.68 396.1 22.0 -33.3 -7.8 60 %2 0.0 0.500 0.0893 1.2597 182.1 72.18 429.4 23.8 - - 40 % 0.0 0.400 0.0893 1.2597 145.7 58.47 470.6 26.1 41.2 9.6 Clay loam 0.095 0.41 0.019 1.31 6.24 76.83 442.9 24.6 - - Silt loam 0.067 0.45 0.02 1.41 10.8 70.85 398.5 22.1 -44.4 -10.0 Loam 0.078 0.43 0.036 1.56 24.96 57.42 479.9 26.6 37.0 8.4 1 Soil water storage at field capacity in 2.6 m deep. 2 Substrate 1 - calibrated model

0.6 0.6 ) ) 3 80% fine particles 100% fine particles 3

/cm 60% fine particles /cm

3 0.5 80% fine particles 0.5 3 60% fine particles 40% fine particles 0.4 0.4

0.3 0.3

0.2 0.2 Volumetric (cm content Volumetric water 0.1 (cm content Volumetric water 0.1 (a) (b)

0.0 0.0 1 100 10000 1000000 1 100 10000 1000000 Matric potential (hPa, log. scale) Matric potential (hPa, log. scale)

Figure 4-23: (a) Water retention curve using 60, 80 and 100 % of fine particles using texture and bulk densities measurements at the Rosetta pedotransfer function; (b) water retention curve according to the relative stone content for substrate 1 The different approaches used to study the use of fine fractions, including bulk densities, soil textures and relative stone content, indicated that lower seepage rates of evapotranspiration covers may be obtained by increasing either the rate of fine particles, < 2 mm, or the clay and silt-size particles in the technogenic substrates. Root development, i.e., root length and root depth, is also expected to improve by increasing the fine fractions rate (Babalola and Lal, 1977) and lead to further reduction of the seepage. Fine particles can additionally be used as interlayers in evapotranspiration covers, as it is recommended in capillary barriers (Radcliffe and Simunek, 2010; Ng et al., 2015). 142

Chapter 4

4.7 Conclusions The aim of the present study was to calibrate the Hydrus-1D mathematical model to make predictions on evapotranspiration and seepage from potash tailings covers. The observed hydraulic properties, seepage and water content of four different technogenic substrates provided the inputs to optimize the van Genuchten water retention parameters. The substrates presented similar pH values, a mean of 8.5, and electrical conductivity with a mean of 3.0 mS/cm. The inverse simulation using weekly measurements of the seepage provided lower differences between predicted and measured drainage and water content values. The calibrated Hydrus-1D model showed a mean variation between observed and predicted values of 2.9 % or circa 13.6 mm for the accumulated seepage of substrates 1-4 from 2014 to 2015, which is lower than the direct simulation, circa 16 % or 75.4 mm. Moreover, the inverse simulation using surface water content and water retention curve measurements deviated from the observed seepage. This indicates the need of having lysimeter outflow measurements to calibrate Hydrus-1D model to simulate water fluxes in evapotranspiration covers. The validation of the Hydrus-1D indicated a seepage rate of 24.7 % and 75.3 % of evapotranspiration for a historical 27-years daily data, which agree with the measurements and simulations presented in the literature. Lower seepage was estimated when increasing the root depth and crop height. The seepage ranged from 20.7 to 9.9 % when using 30 and 100 cm root depth and from 23.7 to 12.9 % of the precipitation when the crop height varied from 30 to 100 cm. Likewise, an increase in the fine particles, < 2 mm diameter, in the substrates may provide further seepage reduction. Overall this study showed that the calibrated Hydrus-1D model reproduced reliable seepage and evapotranspiration values of the potash tailings covers. In addition, the capability of the Hydrus-1D to reproduce the seepage values demonstrates the measurements in the field were properly done. Future studies can consider the temporal settling and compaction of the substrates due to the saturation and desaturation processes. These phenomena may provide distinct values for the hydraulic parameters from the technogenic substrates and seepage rates. Further research can also determine the water fluxes of evapotranspiration covers under different degrees of slope using two or three-dimensional models. Moreover, a longer-term observation of the seepage from substrates would allow one to validate the calibrated model with the experimental data.

143

Chapter 4

4.8 References Abichou, T., Musagasa, J., Tawfiq, K., Chanton, J., 2010. Lysimtery versus deep water content monitoring: Field performance evaluation of alternative landfill covers. GeoFlorida 2010, 2821- 2829. Doi: 10.1061/41095(365)287. Aboukhaled, A., Alfaro, A., Smith, M., 1982. Lysimeters, FAO Irrigation and Drainage Paper 39, Rome. Abtew, W., Melesse, A.M., 2013. Evaporation and evapotranspiration. Measurements and Estimations. London: Springer. AG Boden. 2005. Bodenkundliche Kartieranleitung. (5. Aufl.). Hannover: Bundesanstalt für Geowissenschaften und Rohstoffe. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration - Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, Rome. Babalola, O., Lal, R., 1977. Subsoil gravel horizon and maize root growth. Plant Soil 46, 347-357. Doi: 10.1007/BF00010091. Beckers, E., Pichault, M., Pansak, W., Degré, A., Garré, S., 2016. Characterization of stony soils' hydraulic conductivity using laboratory and numerical experiments. Soil 2, 421-431. Doi: 10.5194/soil-2-421-2016. Bethune, M.G., Selle, B., Wang, Q.J., 2008. Understanding and predicting deep percolation under surface irrigation. Water Resour. Res. 44 (W12430), W12430. Doi: 10.1029/2007WR006380. Bilibio, C., Schellert, C., Retz, S., Hensel, O., Schmeisky, H., Uteau, D., Peth, S., 2017. Water balance assessment of different substrates on potash tailings piles using non-weighable lysimeters. J. Environ. Manage. 196, 633-643. Doi: 10.1016/j.jenvman.2017.01.024. Blume, H.P., Brümmer, G.W., Fleige, H., Horn, R., Kandeler, E., Kögel-Knabner, I., et al., 2016. Scheffer/Schachtschabel Soil Science. (1 ed.). Berlin: Springer. Bouwer, H., Rice, R.C., 1984. Hydraulic properties of stony vadose zones. Groundwater 22, 696- 705. Doi: 10.1111/j.1745-6584.1984.tb01438.x. Brady, N.C., Weil, R.R., 2014. The nature and properties of soils. (14 ed.). Harlow: Pearson. Brakensiek, D.L., Rawls, W.J., Stephenson, G.R., 1986. Determining the saturated hydraulic conductivity of a soil containing rock fragments. Soil Sci. Soc. Am. J. 50, 834. Doi: 10.2136/sssaj1986.03615995005000030053x. Braunisch, F., 2008. Untersuchungen zum Aufbau einer funktional optimierten Rekultivierungsschicht auf einer hochbasischen Aschendeponie (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 30/2008, Universität Kassel, Witzenhausen. Bush, T., 2002. Plant guide for Kentucky Bluegrass (Poa pratensis L.). USDA-Natural Resources Conservation Service. https://plants.usda.gov/factsheet/pdf/fs_popr.pdf (accessed 13 December 2017). 144

Chapter 4

Cañedo-Argüelles, M., Brucet, S., Carrasco, S., Flor-Arnau, N., Ordeix, M., Ponsá, S., Coring, E., 2017. Effects of potash mining on river ecosystems: An experimental study. Environ. Pollut. 224, 759-770. Doi: 10.1016/j.envpol.2016.12.072. Chai, T., Draxler, R.R., 2014. Root mean square error (RMSE) or mean absolute error (MAE)? – Arguments against avoiding RMSE in the literature. Geosci. Model Dev. 7, 1247-1250. Doi: 10.5194/gmd-7-1247-2014. Chang, J., 2009. Climate and agriculture: An ecological survey. London: Aldine Transaction. Couto, M.F., Peternelli, L.A., Barbosa, M.H.P., 2013. Classification of the coefficients of variation for sugarcane crops. Ciência Rural 43, 957-961. Doi: 10.1590/S0103-84782013000600003. Crawley, M.J., 2014. Statistics. An introduction using R. (2 ed.). Wiley. Delta-T Devices, 2013. User manual for the ML3 ThetaProbe soil moisture sensor. Cambridge: Delta-T Devices. Deutsche Wetterdienst, 2017a. DWD Climate Data Center (CDC). ftp://ftp-cdc.dwd.de/pub/CDC/ (accessed 14 August 2017). Deutsche Wetterdienst, 2017b. WEtterdaten und -STatistiken Express (WESTE). http://www.dwd.de/DE/klimaumwelt/cdc/klinfo_systeme/weste/weste_node.html (accessed 01 August 2017). DIN 18124. 2011. Soil, investigation and testing. Determination of density of solid particles (Edition: April 2011). DIN 19683-9. 2012. Soil quality – Physical laboratory tests – Part 9: Determination of the saturated hydraulic water conductivity in the cylindrical core-cutter (Edition: July 2012). DIN ISO 10390. 2005. Soil quality. Determination of pH (Edition: December 2005). DIN ISO 11265. 1997. Soil quality. Determination of the specific electrical conductivity (Edition: June 1997). DIN ISO 11277. 2002. Soil quality. Determination of particle size distribution in mineral soil material. Method by sieving and sedimentation (Edition: August 2002). Doorenbos, J., Pruitt, W.O., 1977. Guidelines for predicting crop water requirements. FAO Irrigation and Drainage Paper 24, Rome.

Food and Agriculture Organization of the United Nations (FAO), 2014. ET0 calculator. Water Development and Management Unit, Information Resources. http://www.fao.org/land- water/databases-and-software/eto-calculator/en/ (accessed 17 January 2016). Field, A.P., 2013. Discovering Statistics using IBM SPSS Statistics. (4 ed.). London. SAGE. Field, A.P., Miles, J., Field, Z., 2012. Discovering statistics using R. London: SAGE. Fiès, J.C., Louvigny, N.D.E., Chanzy, A., 2002. The role of stones in soil water retention. Eur. J. Soil Sci. 53, 95-104. Doi: 10.1046/j.1365-2389.2002.00431.x.

145

Chapter 4

Galleguillos, M., Jacob, F., Prévot, L., Faúndez, C., Bsaibes, A., 2017. Estimation of actual evapotranspiration over a rainfed vineyard using a 1-D water transfer model: A case study within a Mediterranean watershed. Agr. Water Manage. 184, 67-76. Doi: 10.1016/j.agwat.2017.01.006. Geiger, R., Aron, R.H., Todhunter, P., 2009. The climate near the ground. (7 ed.). Lanham: Rowman and Littlefield. Goyal, M.R., Harmsen, E.W., 2014. Evapotranspiration. Principles and applications for water management. Oakville: Apple Academic Press. Hannaway, D., Fransen, S., Cropper, J., Teel, M., Chaney, M., Griggs, T., Halse, R., Hart, J., Cheeke, P., Hansen, D., Klinger, R., Lane, W., 1999. Perennial ryegrass (Lolium perenne L.). Oregon State University Extension Publication PNW503. Hauser, V.L., 2009. Evapotranspiration covers for landfills and waste sites. Boca Raton: CRC Press. Hendriks, M.R., 2010. Introduction to physical hydrology. Oxford: Oxford University Press. Hermsmeyer, D., 2001. Soil physical and hydrological evaluation of aluminum recycling by- product as an infiltration barrier for potash mine tailings (Doctoral Dissertation). Hanover University, Welfengarten. Hlaváčiková, H., Novák, V., 2014. A relatively simple scaling method for describing the unsaturated hydraulic functions of stony soils. J. Plant Nutr. Soil Sci. 177, 560-565. Doi: 10.1002/jpln.201300524. Hoffman, G.J., et al., 2007. Design and operation of farm irrigation systems. (2 ed.). American Society of Agricultural and Biological Engineers. St. Joseph: Michigan. Hopmans, J.W., 2010. Infiltration and unsaturated zone. In Wilderer, P.A. (Ed.). Treatise Water Sci. 103–114. Burlington: Elsevier Science. Horton, R., Horn, R., Bachmann, J., Peth, S., 2016. Essential soil physics. An introduction to soil processes, functions, structure and mechanics. Stuttgart: Schweizerbart Science Publishers. Howard, J., 2017. Anthropogenic soils. Progress in Soil Science. Cham: Springer International Publishing. Kodešová, R., Fér, M., Klement, A., Nikodem, A., Tepl, D., Neuberger, P., Bure, P., 2014. Impact of various surface covers on water and thermal regime of technosol. J. Hydrol. 519, 2272-2288. Doi: 10.1016/j.jhydrol.2014.10.035. Lal, R., Shukla, M., 2004. Principles of soil physics. New York: Taylor and Francis. Lamb, E., 2015. A complete study of evapotranspiration. New York: Callisto Reference. Lamprecht, C., 2017. Heringen - Climate normals. Meteostat. https://www.meteostat.net/climate/heringen-hesse (accessed 28 August 2017). Landesbetrieb Landwirtschaft Hessen, 2017. Wetter. https://www.llh.hessen.de/pflanze/wetter/ (accessed 01 August 2017).

146

Chapter 4

Li, Y., Simunek, J., Jing, L., Zhang, Z., Ni, L., 2014. Evaluation of water movement and water losses in a direct-seeded-rice field experiment using Hydrus-1D. Agr. Water Manag. 142, 38- 46. Doi: 10.1016/j.agwat.2014.04.021. Ludwig, B., 2015. Statistische Auswertungen in bodenkundlich-pflanzenbaulichen Studien. Class notes. Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of unsaturated porous media. Water Resour. Res. 12, 513-522. Doi: 10.1029/WR012i003p00513. Munsell Color Company, 1994. Munsell soil color charts. Revised edition. New Windsor: Munsell Color. Ng, C.W.W., Liu, J., Chen, R., Xu, J., 2015. Physical and numerical modeling of an inclined three- layer (silt/gravelly sand/clay) capillary barrier cover system under extreme rainfall. Waste manage. 38, 210-221. Doi: 10.1016/j.wasman.2014.12.013. Niessing, S., 2005. Begrünungsmaßnahmen auf der Rückstandshalde des Kaliwerkes - Sigmundshall in Bokeloh (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 25/2005, Universität Kassel, Witzenhausen. Novák, V., 2012. Evapotranspiration in the soil-plant-atmosphere system. Progress in Soil Science. London: Springer. Novák, V., Kňava, K., Šimůnek, J., 2011. Determining the influence of stones on hydraulic conductivity of saturated soils using numerical method. Geoderma 161, 177-181. Doi: 10.1016/j.geoderma.2010.12.016. Papke, G., Schmeisky, H., 2013. Rekultivierung von Rückstandshalden der Kaliindustrie. Ergebnisse aus langjährigen wissenschaftlichen Begleituntersuchungen der Begrünungsflächen auf der Kalirückstandshalde Sigmundshall in Bokeloh. Ökologie und Umweltsicherung, Bd. 35/2013, Universität Kassel, Witzenhausen. Papke, G., Schmeisky, H., 2017. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet – Stufe II Feldversuch Lysimeterfeld auf der Halde IV in Heringen - Endbericht Teilbericht A. Umweltsicherung Schmeisky (unveröffentlichter Bericht). Pereira, L.S., Allen, R.G., Smith, M., Raes, D., 2015. Crop evapotranspiration estimation with FAO56. Past and future. Agr. Water Manage. 147, 4-20. Doi: 10.1016/j.agwat.2014.07.031. Podlacha, G., 1999. Untersuchungen zur Substratandeckung mit geringen Schichtstärken aus Bodenaushub-Wirbelschichtasche-Gemischen und ihrer Begrünung (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 16/1999, Universität Kassel, Witzenhausen. Radcliffe, D.E., Simunek, J., 2010. Soil physics with HYDRUS. Modeling and applications. Florida: CRC Press. Rassam, D., Simunek, J., van Genuchten, M. Th., 2003. Modelling variably saturated flow with HYDRUS-2D. Brisbane: ND Consult. 147

Chapter 4

Rauche, H., 2015. Die Kaliindustrie im 21. Jahrhundert. Stand der Technik bei der Rohstoffgewinnung und der Rohstoffaufbereitung sowie bei der Entsorgung der dabei anfallenden Rückstände. (1. Aufl.). Berlin: Springer. Revelle, W., 2015. psych. Procedures for personality and psychological research. Evanston: Northwestern University. http://CRAN.R-project.org/package=psych, updated on 30/8/2015 (accessed 02 September 2016). Richter, D., 1995. Ergebnisse methodischer Untersuchungen zur Korrektur des systematischen Meßfehlers des Hellmann-Niederschlagsmessers. Offenbach am Main, Selbstverl: Deutscher Wetterdienst. Rock, S., Myers, B., Fiedler, L., 2012. Evapotranspiration (ET) covers. Int. J. Phytorem. 14, 1-25. Doi: 10.1080/15226514.2011.609195. RStudio Team. 2015. RStudio. Integrated Development for R. RStudio, Inc. Boston, MA. Rücknagel, J., Götze, P., Hofmann, B., Christen, O., Marschall, K., 2013. The influence of soil gravel content on compaction behaviour and pre-compression stress. Geoderma 209-210, 226- 232. Doi: 10.1016/j.geoderma.2013.05.030. Schaap, M.G., Leij, F.J., van Genuchten, M.Th., 2001. Rosetta: A computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. J. Hydrol. 251, 163-176. Doi: 10.1016/S0022-1694(01)00466-8. Schmeisky, H., Papke, G., 2013. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet – Stufe II Feldversuch Lysimeterfeld auf der Halde IV in Heringen - 1. Zwischenbericht Teilbericht A. Umweltsicherung Schmeisky (unveröffentlichter Bericht). Seiler, K.P., Gat, J.R., 2007. Groundwater recharge from run-off, infiltration and percolation. Water Science and Technology Library, v. 55. Dordrecht: Springer. Selker, J.S., Keller, C.K., MacCord, J.T., 1999. Vadose zone processes. Boca Raton: Lewis. Shukla, M., 2014. Soil physics. An introduction. Boca Raton: CRC Press. Silva, E. L. 2005. Fundamentos de irrigação e drenagem. Lavras: UFLA, 89 p. Simões, W.L., Figueirêdo, V.B., Silva, E.L., 2005. Uso do cilindro infiltrômetro único em diferentes solos. Eng. Agríc. 25, 359-366. Doi: 10.1590/S0100-69162005000200009. Simunek, J., Šejna, M., Saito, H., Sakai, M., van Genuchten, M.Th., 2013. The HYDRUS-1D software package for simulating the movement of water, heat, and multiple solutes in variably saturated media. Version 4.17. Department of Environmental Sciences. Riverside: University of California. Simunek, J., Sejna, M., van Genuchten, M. Th., 2014. HYDRUS-1D code for simulating one- dimensional movement of water, heat, and multiple solutes in variably saturated porous media. https://www.pc-progress.com/ (accessed 13 December 2017).

148

Chapter 4

Simunek, J., van Genuchten, M.Th., Šejna, M., 2008. Development and applications of the HYDRUS and STANMOD software packages and related codes. Vadose Zone J. 7, 587-600. Doi: 10.2136/vzj2007.0077. St. John, L., Tilley, D., Hunt, P., Wright, S. 2012. Plant guide for red fescue (Festuca rubra L.). USDA-Natural Resources Conservation Service, Plant Materials Center, Aberdeen, Idaho. Tan, X., Shao, D., Liu, H., 2014. Simulating soil water regime in lowland paddy fields under different water managements using HYDRUS-1D. Agr. Water Manage. 132, 69-78. Doi: 10.1016/j.agwat.2013.10.009. TerraServer, 2016. Aerial Photos & Satellite Images. https://www.terraserver.com/ (accessed 14 June 2016). Thornthwaite, C.W., 1948. An approach toward a rational classification of climate. Geogr. Rev. 38, 55-94. Doi. 10.2307/210739. Tukimat, N.N.A., Harun, S., Shahid, S., 2012. Comparison of different methods in estimating potential evapotranspiration at Muda Irrigation Scheme of Malaysia. J. Agr. Rural Develop. Trop. Subtrop., 113, 77-85. Kassel University Press. van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892-898. van Genuchten, M.Th., 1992. On estimating the hydraulic properties of unsaturated soils. In van Genuchten, M.Th., Leij, F.J., Lund, L.J., (Eds.). Indirect methods for estimating the hydraulic properties of unsaturated soils. 1-14. Proceedings of the International Workshop, Riverside, CA. van Genuchten, M.Th., Leij, F.J., Yates, S.R., 1991. The RETC code for quantifying the hydraulic functions of unsaturated soils. Environmental Protection Agency Report, 600/2-91/065, U.S. Salinity Laboratory, 93.

4.9 Supplementary materials

149

Chapter 4

0.6 ) 3 S2 (0.0 - 0.04 m) /cm 3 0.5 S2 (0.20 - 0.24 m) S2 (0.40 - 0.44 m)

S2 (cm S2 0.4 S2 (0.60 - 0.64 m) - S2 (0.0 - 0.04 m) 0.3 S2 (0.20 - 0.24 m) S2 (0.40 - 0.44 m) 0.2 S2 (0.60 - 0.64 m) 0.1

0.0 Volumetric contentwater Volumetric 1 10 100 1000 10000 100000 100000010000000 Matric potential (hPa, log. scale)

Figure 4-24: (a) Water retention curve from substrate 2 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6)

) 0.6 3 S3 (0.0 - 0.04 m) /cm 3 0.5 S3 (0.20 - 0.24 m) S3 (0.40 - 0.44 m) S3 (cm S3

- 0.4 S3 (0.60 - 0.64 m) S3 (0.0 - 0.04 m) 0.3 S3 (0.20 - 0.24 m) S3 (0.40 - 0.44 m) 0.2 S3 (0.60 - 0.64 m) 0.1

Volumetric water Volumetric content 0.0 1 10 100 1000 10000 100000 100000010000000 Matric potential (hPa, log. scale)

Figure 4-25: (a) Water retention curve from substrate 3 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6)

150

Chapter 4

0.6 T4 (0.0 - 0.04 m) 0.5 T4 (0.20 - 0.24 m) T4 (0.40 - 0.44 m) S4 (cm³/cm³) S4

- 0.4 T4 (0.60 - 0.64 m) T4 (0.0 - 0.04 m) 0.3 T4 (0.20 - 0.24 m) T4 (0.40 - 0.44 m) 0.2 T4 (0.60 - 0.64 m) 0.1

Volumetric water content Volumetric 0.0 1 10 100 1000 10000 100000 100000010000000 Matric potential (hPa, log. scale)

Figure 4-26: (a) Water retention curve from substrate 4 from 0.0-0.64 m depth. Solid lines are the estimated values according to van Genuchten model and the scatter plots are the observed measurements (n=6)

151

Chapter 4

Table 4-7: pH values of substrates 1-4 considering the DIN ISO 10390 in 2014 (n=2) and 2016 (n=3) according to different depths

Substrate Substrate 1 Substrate 2 Substrate 3 Substrate 4 S1-S4 S1-S4 Depth (m) 2014 SD 2016 SD 2014 SD 2016 SD 2014 SD 2016 SD 2014 SD 2016 SD Ø 2014 SD Ø 2016 SD 0.0 - 0.2 8.1 0.0 7.9 0.1 8.1 0.0 8.0 0.1 8.1 0.0 8.0 0.0 8.2 0.0 8.1 0.0 8.1 0.1 8.0 0.1 0.2 - 0.4 8.2 0.0 8.2 0.0 8.4 0.1 8.2 0.0 8.2 0.0 8.2 0.1 8.5 0.1 8.3 0.1 8.3 0.1 8.2 0.1 1.4 - 1.6 8.7 0.0 8.9 0.1 8.8 0.1 9.0 0.1 8.9 0.0 8.9 0.1 9.5 0.0 9.5 0.1 8.9 0.4 9.0 0.3 2.0 - 2.2 8.5 0.3 8.8 0.0 8.8 0.2 8.9 0.0 8.9 0.0 9.0 0.0 9.4 0.1 9.5 0.1 8.8 0.4 9.0 0.3 2.4 - 2.6 8.7 0.0 8.9 0.1 8.6 0.0 9.0 0.1 8.9 0.1 9.0 0.2 9.5 n.d. 9.6 0.1 8.8 0.4 9.0 0.3 2.8 - 3.0 8.7 n.d. 9.0 0.1 8.8 0.0 9.0 0.2 8.9 0.1 9.1 0.1 9.4 n.d. 9.4 0.2 8.9 0.3 9.1 0.2 0.0-3.0 Minimum 8.1 7.9 8.1 8.0 8.1 8.0 8.2 8.1 8.1 8.0

Mean 8.4 8.4 8.5 8.5 8.5 8.5 8.7 8.6 8.5 8.5 Maximum 8.7 9.0 8.8 9.0 8.9 9.1 9.5 9.6 8.9 9.1 SD 0.3 0.5 0.3 0.4 0.4 0.5 0.6 0.7 0.3 0.5 CV 3.6 5.5 3.3 5.2 4.4 5.5 6.6 7.8 3.9 5.7

Ø: mean of the substrates 1-4; SD: standard deviation; CV: coefficient of variation; n.d.: not determined

152

Chapter 4

Table 4-8: Electrical conductivity (mS/cm) of substrates 1-4 considering the DIN ISO 11265 in 2014 (n=2) and 2016 (n=3) according to different depths

Substrate Substrate 1 Substrate 2 Substrate 3 Substrate 4 S1-S4 S1-S4 Depth (m) 2014 SD 2016 SD 2014 SD 2016 SD 2014 SD 2016 SD 2014 SD 2016 SD Ø 2014 SD Ø 2016 SD 0.0 - 0.2 2.5 0.1 2.5 0.0 2.5 0.3 2.4 0.4 2.5 0.4 2.4 0.0 2.6 0.1 2.4 0.0 2.6 0.1 2.4 0.0 0.2 - 0.4 2.6 0.1 2.6 0.1 2.6 0.0 2.6 0.0 2.6 0.0 2.5 0.1 2.6 0.0 2.5 0.1 2.6 0.0 2.5 0.1 1.4 - 1.6 3.0 0.4 2.7 0.1 3.5 0.8 2.7 0.1 3.0 0.4 2.6 0.1 3.1 0.6 2.8 0.5 3.1 0.2 2.7 0.1 2.0 - 2.2 4.1 0.4 2.9 0.2 4.2 1.2 2.9 0.2 4.0 0.1 2.7 0.1 3.8 1.1 3.3 1.1 4.0 0.1 3.0 0.3 2.4 - 2.6 4.0 0.1 3.3 0.3 4.4 1.5 2.9 0.1 4.0 0.3 2.9 0.2 4.4 n.d. 3.5 1.3 4.2 0.2 3.2 0.3 2.8 - 3.0 5.3 n.d. 6.0 1.7 5.6 1.3 15.0 20.2 9.1 3.9 5.8 2.6 5.1 n.d. 3.6 0.3 6.2 1.9 7.6 5.0 0.0-3.0 Minimum 2.5 2.5 2.6 2.4 2.5 2.4 2.6 2.4 2.6 2.4

Mean 3.6 3.3 3.8 4.7 4.2 3.2 3.6 3.0 3.8 3.6 Maximum 5.3 6.0 5.6 15.0 9.1 5.8 5.1 3.6 6.2 7.6 SD 1.1 1.3 1.1 5.0 2.5 1.3 1.0 0.5 1.4 2.0 CV 30.3 40.3 29.5 105.7 59.6 41.3 27.8 18.1 36.5 55.9

Ø: mean of the substrates 1-4; SD: standard deviation; CV: coefficient of variation; n.d.: not determined

153

Chapter 4

Table 4-9: Bulk density (g/cm³) of substrates 1-4 2016 considering different depths (n=6)

Substrate Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1-4 Depth (m) Mean SD Mean SD Mean SD Mean SD Ø SD 0.0 - 0.04 1.16 0.02 1.15 0.03 1.18 0.01 1.16 0.06 1.16 0.01 0.20 - 0.24 1.17 0.01 1.21 0.03 1.24 0.02 1.13 0.01 1.19 0.05 0.40 - 0.44 1.25 0.04 1.22 0.03 1.32 0.02 1.20 0.04 1.25 0.05 0.60 - 0.64 1.23 0.03 1.29 0.03 1.28 0.02 1.17 0.04 1.24 0.05 Minimum 1.16 1.15 1.18 1.13 1.16 Mean 1.20 1.21 1.25 1.17 1.21 Maximum 1.25 1.29 1.32 1.20 1.25 SD 0.04 0.06 0.06 0.03 0.04 CV 3.6 4.6 4.7 2.7 3.4

Mean: arithmetic mean of 6 repetitions; Ø: mean of the substrates 1-4; SD: standard deviation; CV: coefficient of variation

154

Chapter 4

Table 4-10: Estimated total porosity (%) in substrates 1-4 2016 considering different depths (n=6)

Substrate Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1-4 Depth (m) Mean SD Mean SD Mean SD Mean SD Ø SD 0.0 - 0.04 54.9 1.0 55.1 1.0 53.9 0.4 54.7 2.4 54.7 0.5 0.20 - 0.24 54.3 0.6 52.8 1.2 51.7 0.7 56.0 0.4 53.7 1.9 0.40 - 0.44 51.4 1.6 52.5 1.3 48.5 0.9 53.0 1.4 51.4 2.0 0.60 - 0.64 52.0 1.3 49.8 1.3 50.1 0.8 54.3 1.4 51.6 2.1 Minimum 51.4 49.8 48.5 53.0 51.4 Mean 53.1 52.5 51.0 54.5 52.8 Maximum 54.9 55.1 53.9 56.0 54.7 SD 1.70 2.2 2.3 1.2 1.6 CV 3.2 4.2 4.5 2.3 3.1

Mean: arithmetic mean of 6 repetitions; Ø: mean of the substrates 1-4; SD: standard deviation; CV: coefficient of variation

155

Chapter 4

Table 4-11: Available and unavailable moisture in the substrates 1-4 (n=6)

Substrate Water content at saturation (%) % drainage % available moisture % not available moisture Depth (m) S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4 S1 S2 S3 S4 0.0 - 0.04 52.4 52.6 50.2 48.8 17.3 15.0 15.1 14.9 23.4 23.6 23.2 20.1 11.7 14.1 11.9 13.8 0.20 - 0.24 51.8 48.9 48.0 48.9 17.8 13.9 14.2 17.6 22.1 20.2 21.4 18.4 11.8 14.8 12.4 12.9 0.40 - 0.44 48.1 48.2 44.1 48.1 15.2 15.0 11.1 14.9 21.2 19.4 20.8 18.4 11.7 13.8 12.2 14.8 0.60 - 0.64 48.6 45.8 44.6 47.7 15.2 13.0 12.4 15.8 22.0 19.5 20.0 16.9 11.4 13.3 12.2 15.0 Minimum 48.1 45.8 44.1 47.7 15.2 13.0 11.1 14.9 21.2 19.4 20.0 16.9 11.4 13.3 11.9 12.9 Mean 50.2 48.9 46.7 48.4 16.4 14.2 13.2 15.8 22.2 20.7 21.3 18.4 11.7 14.0 12.2 14.1 Maximum 52.4 52.6 50.2 48.9 17.8 15.0 15.1 17.6 23.4 23.6 23.2 20.1 11.8 14.8 12.4 15.0 SD 2.2 2.8 2.9 0.6 1.4 1.0 1.8 1.3 0.9 2.0 1.3 1.3 0.2 0.6 0.2 1.0 CV 4.3 5.8 6.2 1.2 8.5 6.9 13.6 8.1 4.2 9.5 6.3 7.1 1.5 4.5 1.7 6.9

S: substrate; SD: standard deviation; CV: coefficient of variation

156

Chapter 4

Table 4-12: Pore size distribution of the substrates 1-4 (n=24)

Pore size Diameter Substrate 1 Substrate 2 Substrate 3 Substrate 4 Mean S1-S4 SD CV

distribution µm hPa ------Volume-%------Volume-%--- % Coarse pores >50 < 60 16.4 14.2 13.2 15.8 14.9 1.5 9.8 Tight coarse pores 50 to >10 60 - 300 7.1 6.6 5.8 7.5 6.7 0.7 10.8 Medium pores 10 to > 0.2 300 - < 15000 15.1 14.0 15.6 11.0 13.9 2.1 14.8 Fine pores ≤ 0.2 ≥ 15000 11.7 14.0 12.2 14.1 13.0 1.3 9.7

SD: standard deviation; CV: coefficient of variation

157

Chapter 4

Table 4-13: Saturated hydraulic conductivity in substrates 1-4 2016 considering different depths (n=6)

Substrate Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1-4 Depth (m) Mean SD Mean SD Mean SD Mean SD Ø SD 0.0 - 0.04 785.5 199.5 888.8 391.3 474.6 105.7 844.2 284.8 748.3 187.3 0.20 - 0.24 902.7 171.8 754.3 164.4 902.7 171.8 1098.7 212.1 914.6 141.3 0.40 - 0.44 546.9 163.1 514.2 232.4 178.0 85.5 574.0 262.7 453.3 185.2 0.60 - 0.64 514.5 95.8 138.7 15.0 296.5 112.3 270.9 38.9 305.1 155.7 Minimum 514.5 138.7 178.0 270.9 305.1 Mean 687.4 574.0 462.9 697.0 605.3 Maximum 902.7 888.8 902.7 1098.7 914.6 SD 187.6 329.0 317.5 355.8 276.5 CV 27.3 57.3 68.6 51.0 45.7

Mean: arithmetic mean of 6 repetitions; Ø: mean of the substrates 1-4; SD: standard deviation; CV: coefficient of variation

158

Chapter 4

Table 4-14: Particle size distribution of substrates 1-4 (n=12)

Coarse fraction Fine fraction (%) Classification of the fine fraction Substrates Mean SD Mean Sand SD Mean Silt SD Mean Clay SD

Substrate 1 48.8 6.0 54.8 4.2 41.3 4.7 3.9 4.1 Sandy loam

Substrate 2 39.8 7.0 47.6 2.9 47.4 4.4 5.0 4.0 Silt loam

Substrate 3 38.1 5.0 51.2 4.6 42.9 5.6 5.9 3.1 Sandy loam

Substrate 4 40.6 5.8 54.1 5.7 38.9 3.4 7.0 3.8 Sandy loam

Minimum 38.1 47.6 38.9 3.9

Mean 41.8 51.9 42.6 5.5 Sandy loam Maximum 48.8 54.8 47.4 7.0 SD 4.8 3.3 3.6 1.3 CV 11.4 6.3 8.4 24.2 Mean: arithmetic mean of 12 repetitions; SD: standard deviation; CV: coefficient of variation

159

Chapter 4

Table 4-15: Root-mean-square error (RMSE) and correlation coefficients between observed and forward simulations of seepage and water content using Hydrus-1D from substrates 1-4

Seepage Water content Substrates RMSE Correlation RMSE Correlation mm coefficient p-value R2 cm³/cm³ coefficient p-value R2 Substrate 1 34.8 0.99 0.0 0.98 0.063 0.80 0.0 0.64 Substrate 2 49.9 1.0 0.0 1.0 0.086 0.71 0.0 0.50 Substrate 3 52.2 0.98 0.0 0.96 0.064 0.74 0.0 0.55 Substrate 4 49.4 1.0 0.0 1.0 0.068 0.73 0.0 0.53

Table 4-16: Observed and fitted hydraulic properties of substrate 1 considering five different input data in the inverse solution of Hydrus-1D

2 Inverse Ɵs α n Ks R Iteration Error RMSE input data time Seepage WC cm³/cm³ 1/cm cm/d - nr. % sec mm cm³/cm³ Observed 0.5038 0.0777 1.2103 687.40 0.9933 Inverse 1 0.5000 0.0893 1.2597 182.14 0.9974 7 27.3 0.2409 12.1 0.065 Inverse 2 0.5000 0.1141 1.2633 319.27 1.0000 6 25.4 0.4136 13.4 0.059 Inverse 3 0.4760 0.0382 1.2649 1000 0.9071 6 19.0 0.1127 75.9 0.053 Inverse 4 0.4940 0.1122 1.2681 301.46 0.9999 7 28.8 0.4069 13.6 0.060 Inverse 5 0.4941 0.0796 1.2892 124.15 0.9987 8 34.8 0.3952 12.0 0.062 Inverse data 1: seepage (104), water content (1); Inverse data 2: seepage (4), water content (1); Inverse data 3: water content (19), water retention curve (1); Inverse data 4: seepage (4); water content (19); Inverse data 5: seepage (24), water content (19). Observed: mean hydraulic parameters from 0-0.64 m depth, Inverse: hydraulic parameters from 0.0-2.60 m depth. RMSE: root-mean-square error; WC: water content

160

Chapter 4

Table 4-17: Observed and fitted hydraulic properties of substrate2 considering five different inputs in the inverse solution of Hydrus-1D

2 Inverse Ɵs α n Ks R Iteration Error RMSE input data time Seepage WC cm³/cm³ 1/cm cm/d - nr. % sec mm cm³/cm³ Observed 0.4879 0.0696 1.1913 574.00 0.9910 Inverse 1 0.5000 0.0721 1.3428 89.19 0.9969 13 51.9 0.3897 14.2 0.058 Inverse 2 0.5000 0.1502 1.3204 511.61 0.9994 12 50.5 0.4356 16.9 0.058 Inverse 3 0.4157 0.0076 1.3694 66.30 0.9176 20 53.9 0.6062 163.1 0.040 Inverse 4 0.4841 0.0823 1.3753 115.25 0.9995 6 32.0 0.3370 14.9 0.061 Inverse 5 0.5000 0.1057 1.3328 219.57 0.9979 4 18.0 0.3983 15.1 0.057 Inverse data 1: seepage (104), water content (1); Inverse data 2: seepage (4), water content (1); Inverse data 3: water content (19), water retention curve (1); Inverse data 4: seepage (4); water content (19); Inverse data 5: seepage (24), water content (19). Observed: mean hydraulic parameters from 0-0.64 m depth. Inverse: hydraulic parameters from 0.0-2.60 m depth. RMSE: root-mean-square error; WC: water content

Table 4-18: Observed and fitted hydraulic properties of substrate 3 considering five different inputs in the inverse solution of Hydrus-1D

2 Inverse Ɵs α n Ks R Iteration Error RMSE input data time Seepage WC cm³/cm³ 1/cm cm/d - nr. % sec mm cm³/cm³ Observed 0.4736 0.0660 1.1850 463.0 0.9867 Inverse 1 0.5000 0.1257 1.3382 347.2 0.9959 8 33.5 0.3975 18.4 0.079 Inverse 2 0.5000 0.1861 1.3666 938.0 0.9992 4 19.1 0.5567 18.5 0.101 Inverse 3 0.4943 0.0683 1.2196 1000.0 0.8197 3 18.4 0.3567 51.5 0.056 Inverse 4 0.5000 0.1775 1.3663 835.4 0.9992 6 26.8 0.5214 18.4 0.099 Inverse 5 0.5000 0.1008 1.3947 183.2 0.9977 9 35.5 0.4307 17.3 0.086 Inverse data 1: seepage (104), water content (1); Inverse data 2: seepage (4), water content (1); Inverse data 3: water content (19), water retention curve (1); Inverse data 4: seepage (4); water content (19); Inverse data 5: seepage (24), water content (19). Observed: mean hydraulic parameters from 0-0.64 m depth. Inverse: hydraulic parameters from 0.0-2.60 m depth. RMSE: root-mean-square error; WC: water content

161

Chapter 4

Table 4-19: Observed and fitted hydraulic properties of substrate 4 considering five different inputs in the inverse solution of Hydrus-1D

2 Inverse Ɵs α n Ks R Iteration Error RMSE input data time Seepage WC cm³/cm³ 1/cm cm/d - nr. % sec mm cm³/cm³ Observed 0.4913 0.1025 1.1946 697.00 0.9896 Inverse 1 0.5000 0.1537 1.3509 526.47 0.9955 8 33.6 0.5450 20.5 0.091 Inverse 2 0.5000 0.1766 1.4789 691.16 0.9993 10 43.8 0.7003 17.5 0.127 Inverse 3 0.5000 0.0803 1.2299 374.13 0.8006 6 33.1 0.3527 35.1 0.061 Inverse 4 0.4990 0.1625 1.4408 577.61 0.9990 6 29.0 0.6174 18.0 0.116 Inverse 5 0.5000 0.1771 1.3715 765.23 0.9970 9 40.9 0.5134 19.0 0.102 Inverse data 1: seepage (104), water content (1); Inverse data 2: seepage (4), water content (1); Inverse data 3: water content (19), water retention curve (1); Inverse data 4: seepage (4); water content (19); Inverse data 5: seepage (24), water content (19). Observed: mean hydraulic parameters from 0-0.64 m depth. Inverse: hydraulic parameters from 0.0-2.60 m depth. RMSE: root-mean-square error; WC: water content

162

Chapter 4

Table 4-20: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly accumulated seepage (104) and one water content measurement (1)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 265.2 303.9 260.8 38.7 14.6 -4.4 -1.6 2015 173.9 112.4 172.4 -61.5 -35.4 -1.5 -0.9 Total 439.1 416.1 433.3 -23.0 -5.2 -5.8 -1.3 Substrate 2 2014 261.2 232.0 261.7 -29.2 -11.2 0.5 0.2 2015 197.4 132.9 188.5 -64.5 -32.7 -8.9 -4.5 Total 458.6 364.9 450.2 -93.7 -20.4 -8.4 -1.8 Substrate 3 2014 275.4 269.4 275.5 -6.0 -2.2 0.1 0.0 2015 196.9 108.6 181.1 -88.3 -44.9 -15.8 -8.0 Total 472.3 378.0 456.6 -94.3 -20.0 -15.7 -3.3 Substrate 4 2014 282.9 245.8 277.8 -37.1 -13.1 -5.1 -1.8 2015 200.2 146.8 180.7 -53.4 -26.7 -19.5 -9.7 Total 483.1 392.6 458.5 -90.5 -18.7 -24.6 -5.1

163

Chapter 4

Table 4-21: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly accumulated seepage (104) and one water content measurement (1)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 523.0 535.1 511.2 12.1 2.3 -11.8 -2.3 2015 369.9 451.0 435.6 81.1 21.9 65.7 17.8 Total 892.9 986.1 946.8 93.2 10.4 53.9 6.0 Substrate 2 2014 527.1 557.3 512.3 30.2 5.7 -14.9 -2.8 2015 346.4 463.6 436.1 117.2 33.8 89.7 25.9 Total 873.5 1020.8 948.4 147.3 16.9 74.9 8.6 Substrate 3 2014 512.9 557.9 497.9 45.0 8.8 -15.1 -2.9 2015 346.9 470.6 425.7 123.7 35.7 78.8 22.7 Total 859.8 1028.5 923.6 168.7 19.6 63.8 7.4 Substrate 4 2014 505.3 533.9 489.4 28.6 5.7 -15.9 -3.1 2015 343.6 445.7 420.2 102.1 29.7 76.6 22.3 Total 848.9 979.6 909.6 130.7 15.4 60.7 7.2

164

Chapter 4

Table 4-22: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with accumulated seepage by season (4) and one water content measurement (1)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 265.2 303.9 266.2 38.7 14.6 1.0 0.4 2015 173.9 112.4 171.7 -61.5 -35.4 -2.2 -1.3 Total 439.1 416.1 437.9 -23.0 -5.2 -1.2 -0.3 Substrate 2 2014 261.2 232.0 237.6 -29.2 -11.2 -23.7 -9.1 2015 197.4 132.9 178.9 -64.5 -32.7 -18.5 -9.4 Total 458.6 364.9 452.4 -93.7 -20.4 -6.2 -1.3 Substrate 3 2014 275.4 269.4 289.2 -6.0 -2.2 13.8 5.0 2015 196.9 108.6 175.3 -88.3 -44.9 -21.6 -11.0 Total 472.3 378.0 464.5 -94.3 -20.0 -7.8 -1.7 Substrate 4 2014 282.9 245.8 291.1 -37.1 -13.1 8.2 2.9 2015 200.2 146.8 184.9 -53.4 -26.7 -15.3 -7.7 Total 483.1 392.6 475.9 -90.5 -18.7 -7.2 -1.5

165

Chapter 4

Table 4-23: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with accumulated seepage by season (4) and water content measurement (1)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 523.0 535.1 507.6 12.1 2.3 -15.4 -2.9 2015 369.9 451.0 430.9 81.1 21.9 61.0 16.5 Total 892.9 986.1 938.6 93.2 10.4 45.7 5.1 Substrate 2 2014 527.1 557.3 494.5 30.2 5.7 -32.6 -6.2 2015 346.4 463.6 423.2 117.2 33.8 76.8 22.2 Total 873.5 1020.8 917.8 147.3 16.9 44.3 5.1 Substrate 3 2014 512.9 557.9 482.6 45.0 8.8 -30.3 -5.9 2015 346.9 470.6 413.6 123.7 35.7 66.7 19.2 Total 859.8 1028.5 896.3 168.7 19.6 36.5 4.2 Substrate 4 2014 505.3 533.9 466.7 28.6 5.7 -38.6 -7.6 2015 343.6 445.7 402.2 102.1 29.7 58.6 17.1 Total 848.9 979.6 869.0 130.7 15.4 20.1 2.4

166

Chapter 4

Table 4-24: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and retention curve, ψm(Ɵ) (1)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 265.2 303.9 250.0 38.7 14.6 -15.2 -5.7 2015 173.9 112.4 71.7 -61.5 -35.4 -102.2 -58.8 Total 439.1 416.1 321.7 -23.0 -5.2 -117.4 -26.7 Substrate 2 2014 261.2 232.0 178.8 -29.2 -11.2 -82.4 -31.5 2015 197.4 132.9 3.0 -64.5 -32.7 -194.4 -98.5 Total 458.6 364.9 181.8 -93.7 -20.4 -276.8 -60.4 Substrate 3 2014 275.4 269.4 304.7 -6.0 -2.2 29.3 10.6 2015 196.9 108.6 97.1 -88.3 -44.9 -99.8 -50.7 Total 472.3 378.0 401.8 -94.3 -20.0 -70.5 -14.9 Substrate 4 2014 282.9 245.8 292.6 -37.1 -13.1 9.7 3.4 2015 200.2 146.8 134.3 -53.4 -26.7 -65.9 -32.9 Total 483.1 392.6 426.9 -90.5 -18.7 -56.2 -11.6

167

Chapter 4

Table 4-25: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and retention curve, ψm(Ɵ) (1)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 523.0 535.1 663.3 12.1 2.3 140.3 26.8 2015 369.9 451.0 593.5 81.1 21.9 223.6 60.4 Total 892.9 986.1 1256.8 93.2 10.4 363.9 40.8 Substrate 2 2014 527.1 557.3 770.0 30.2 5.7 242.9 46.1 2015 346.4 463.6 732.9 117.2 33.8 386.5 111.6 Total 873.5 1020.8 1502.9 147.3 16.9 629.4 72.1 Substrate 3 2014 512.9 557.9 560.6 45.0 8.8 47.7 9.3 2015 346.9 470.6 478.2 123.7 35.7 131.3 37.8 Total 859.8 1028.5 1038.8 168.7 19.6 179.0 20.8 Substrate 4 2014 505.3 533.9 524.0 28.6 5.7 18.7 3.7 2015 343.6 445.7 441.7 102.1 29.7 98.1 28.6 Total 848.9 979.6 965.6 130.7 15.4 116.7 13.7

168

Chapter 4

Table 4-26: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and seasonal accumulated seepage (4)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 265.2 303.9 266.6 38.7 14.6 1.4 0.5 2015 173.9 112.4 172.0 -61.5 -35.4 -1.9 -1.1 Total 439.1 416.1 438.6 -23.0 -5.2 -0.5 -0.1 Substrate 2 2014 261.2 232.0 268.4 -29.2 -11.2 7.2 2.7 2015 197.4 132.9 189.6 -64.5 -32.7 -7.8 -4.0 Total 458.6 364.9 458.0 -93.7 -20.4 -0.6 -0.1 Substrate 3 2014 275.4 269.4 288.4 -6.0 -2.2 13.0 4.7 2015 196.9 108.6 176.3 -88.3 -44.9 -20.6 -10.5 Total 472.3 378.0 464.7 -94.3 -20.0 -7.6 -1.6 Substrate 4 2014 282.9 245.8 288.4 -37.1 -13.1 5.5 2.0 2015 200.2 146.8 184.0 -53.4 -26.7 -16.2 -8.1 Total 483.1 392.6 472.5 -90.5 -18.7 -10.6 -2.2

169

Chapter 4

Table 4-27: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and seasonal accumulated seepage (4)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 523.0 535.1 506.1 12.1 2.3 -16.9 -3.2 2015 369.9 451.0 432.0 81.1 21.9 62.1 16.8 Total 892.9 986.1 938.1 93.2 10.4 45.2 5.1 Substrate 2 2014 527.1 557.3 504.0 30.2 5.7 -23.1 -4.4 2015 346.4 463.6 429.0 117.2 33.8 82.6 23.8 Total 873.5 1020.8 933.0 147.3 16.9 59.5 6.8 Substrate 3 2014 512.9 557.9 484.4 45.0 8.8 -28.5 -5.6 2015 346.9 470.6 415.2 123.7 35.7 68.3 19.7 Total 859.8 1028.5 899.6 168.7 19.6 39.8 4.6 Substrate 4 2014 505.3 533.9 476.0 28.6 5.7 -29.3 -5.8 2015 343.6 445.7 408.4 102.1 29.7 64.8 18.9 Total 848.9 979.6 884.4 130.7 15.4 35.5 4.2

170

Chapter 4

Table 4-28: Observed and simulated seepage of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and monthly accumulated seepage (24)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 265.2 303.9 259.1 38.7 14.6 -6.1 -2.3 2015 173.9 112.4 179.1 -61.5 -35.4 5.2 3.0 Total 439.1 416.1 438.2 -23.0 -5.2 -0.9 -0.2 Substrate 2 2014 261.2 232.0 267.7 -29.2 -11.2 6.5 2.5 2015 197.4 132.9 185.0 -64.5 -32.7 -12.4 -6.3 Total 458.6 364.9 452.7 -93.7 -20.4 -5.9 -1.3 Substrate 3 2014 275.4 269.4 273.3 -6.0 -2.2 -2.1 -0.8 2015 196.9 108.6 191.4 -88.3 -44.9 -5.6 -2.8 Total 472.3 378.0 464.7 -94.3 -20.0 -7.7 -1.6 Substrate 4 2014 282.9 245.8 284.8 -37.1 -13.1 1.9 0.7 2015 200.2 146.8 178.9 -53.4 -26.7 -21.3 -10.6 Total 483.1 392.6 463.7 -90.5 -18.7 -19.4 -4.0

171

Chapter 4

Table 4-29: Estimated and simulated evapotranspiration of substrates 1-4 in 2014 and 2015 when Hydrus-1D was calibrated with weekly water content measurements (19) and monthly accumulated seepage (24)

Observed Simulated Difference Non- Difference Calibrated Water year Non-calibrated Calibrated Calibrated and Observed and Observed mm mm mm mm % mm % Substrate 1 2014 523.0 535.1 512.3 12.1 2.3 -10.7 -2.0 2015 369.9 451.0 435.6 81.1 21.9 65.7 17.8 Total 892.9 986.1 948.0 93.2 10.4 55.1 6.2 Substrate 2 2014 527.1 557.3 502.8 30.2 5.7 -24.3 -4.6 2015 346.4 463.6 429.2 117.2 33.8 82.8 23.9 Total 873.5 1020.8 932.0 147.3 16.9 58.5 6.7 Substrate 3 2014 512.9 557.9 498.5 45.0 8.8 -14.4 -2.8 2015 346.9 470.6 424.5 123.7 35.7 77.6 22.4 Total 859.8 1028.5 922.9 168.7 19.6 63.1 7.3 Substrate 4 2014 505.3 533.9 483.6 28.6 5.7 -21.7 -4.3 2015 343.6 445.7 414.9 102.1 29.7 71.3 20.7 Total 848.9 979.6 898.5 130.7 15.4 49.6 5.8

172

Chapter 4

Table 4-30: Water fluxes for Bad Hersfeld from 1990 to 2016 using hydraulic parameters from substrate 1

Potential Actual Root Actual root Water Precipi Evapo Seepage Water root water water water Seepage year tation ration rate storage uptake uptake uptake cm cm cm cm % cm % cm 1990 70.3 78.7 0.6 49.5 70.4 27.6 39.3 82.7 1991 50.4 70.5 0.5 39.8 78.8 17.8 35.4 74.8 1992 69.2 75.8 0.5 54.2 78.3 7.2 10.4 81.6 1993 62.3 77.4 0.6 43.3 69.5 17.2 27.6 82.1 1994 87.8 76.5 0.6 57.5 65.5 24.6 28.0 86.0 1995 75.0 74.2 0.6 56.8 75.8 22.5 30.0 81.6 1996 51.7 63.5 0.5 41.6 80.4 6.3 12.2 83.9 1997 54.0 72.0 0.5 47.9 88.6 11.5 21.3 77.1 1998 71.7 75.1 0.6 46.8 65.3 5.7 7.9 95.1 1999 60.8 74.7 0.6 53.3 87.8 24.5 40.3 77.1 2000 75.9 70.4 0.5 50.9 67.1 14.0 18.4 87.4 2001 65.6 73.1 0.5 51.7 78.8 18.0 27.5 81.9 2002 82.0 70.4 0.5 58.2 71.0 16.3 19.9 89.0 2003 58.7 77.7 0.6 38.5 65.6 26.7 45.5 80.7 2004 66.1 72.8 0.5 53.3 80.5 10.2 15.5 81.8 2005 73.3 69.7 0.5 56.6 77.2 13.6 18.5 83.8 2006 67.3 67.5 0.5 52.9 78.6 14.5 21.6 83.0 2007 83.1 77.7 0.6 60.1 72.3 16.6 19.9 87.9 2008 58.9 81.8 0.6 46.9 79.6 17.8 30.1 81.0 2009 62.3 71.1 0.5 51.5 82.7 8.4 13.5 82.6 2010 68.0 74.6 0.6 49.4 72.7 16.0 23.5 85.8 2011 57.4 85.2 0.6 47.2 82.3 17.3 30.2 77.8 2012 65.1 74.7 0.6 54.4 83.5 6.9 10.6 80.3 2013 77.1 71.1 0.5 52.1 67.6 17.2 22.4 88.2 2014 75.0 73.7 0.5 50.6 67.5 22.5 30.0 89.5 2015 54.3 81.2 0.6 43.5 80.1 16.4 30.3 82.6 2016 66.6 73.4 0.5 52.3 78.6 12.1 18.1 84.8 Total 1809.9 2004.3 14.9 1360.8 429.4 Mean 67.0 74.2 0.6 50.4 76 15.9 24.0 83.3 SD 9.7 4.6 0.0 5.6 6.9 6.2 9.6 4.4 CV 14.5 6.2 6.1 11.2 9.0 38.9 40.1 5.3 SD: standard deviation; CV: coefficient of variation

173

Chapter 4

Table 4-31: Components of the simplified water balance equation for Bad Hersfeld from 1990 to 2016 using hydraulic parameters from substrate 1

Water year Precipitation Evapotranspiration Evapotranspiration Seepage Seepage rate

cm cm % cm % 1990 70.3 42.7 60.7 27.6 39.3 1991 50.4 32.6 64.6 17.8 35.4 1992 69.2 62.0 89.6 7.2 10.4 1993 62.3 45.1 72.4 17.2 27.6 1994 87.8 63.2 72.0 24.6 28.0 1995 75.0 52.5 70.0 22.5 30.0 1996 51.7 45.4 87.8 6.3 12.2 1997 54.0 42.5 78.7 11.5 21.3 1998 71.7 66.0 92.1 5.7 7.9 1999 60.8 36.3 59.7 24.5 40.3 2000 75.9 62.0 81.6 14.0 18.4 2001 65.6 47.6 72.5 18.0 27.5 2002 82.0 65.7 80.1 16.3 19.9 2003 58.7 32.0 54.5 26.7 45.5 2004 66.1 55.9 84.5 10.2 15.5 2005 73.3 59.8 81.5 13.6 18.5 2006 67.3 52.8 78.4 14.5 21.6 2007 83.1 66.6 80.1 16.6 19.9 2008 58.9 41.2 69.9 17.8 30.1 2009 62.3 53.9 86.5 8.4 13.5 2010 68.0 52.0 76.5 16.0 23.5 2011 57.4 40.0 69.8 17.3 30.2 2012 65.1 58.2 89.4 6.9 10.6 2013 77.1 59.8 77.6 17.2 22.4 2014 75.0 52.5 70.0 22.5 30.0 2015 54.3 37.9 69.7 16.4 30.3 2016 66.6 54.5 81.9 12.1 18.1 Total 1809.9 1380.5 429.4 Mean 67.0 51.1 76.0 15.9 24.0 SD 9.7 10.5 9.6 6.2 9.6 CV 14.5 20.5 12.7 38.9 40.1 SD: standard deviation; CV: coefficient of variation

174

Chapter 4

Table 4-32: Hydraulic properties and accumulated seepage of different substrates using 100 % fine particles 2014 to 2015 in Heringen

1 Substrate Bd Ɵr Ɵs α n Ks Volume Seepage g/cm³ cm³/cm³ cm³/cm³ 1/cm cm/d mm mm % Substrate 1 1.22 0.0340 0.3988 0.0149 1.4688 108.44 463.1 311.3 24.2 Substrate 2 1.19 0.0374 0.3998 0.0090 1.5405 97.48 433.6 220.2 17.1 Substrate 3 1.21 0.0377 0.4026 0.0113 1.5060 90.12 458.3 262.0 20.4 Substrate 4 1.14 0.0405 0.4278 0.0120 1.4915 119.56 485.0 259.4 20.2 Mean 1.19 0.0373 0.4066 0.0115 1.5014 101.2 459.6 259.6 20.2 1 Soil water storage at field capacity in 2.6 m deep

Table 4-33: Hydraulic properties and accumulated seepage of different substrates using 80 % fine particles 2014 to 2015 in Heringen

Substrate Bd Ɵr Ɵs α n Ks Volume¹ Seepage Differences g/cm³ cm³/cm³ cm³/cm³ 1/cm cm/d mm mm % mm % Substrate 1 1.29 0.0329 0.3840 0.0168 1.4602 84.96 460.1 355.6 27.6 44.3 14.2 Substrate 2 1.25 0.0362 0.3868 0.0100 1.5272 77.51 434.0 246.0 19.1 25.8 11.7 Substrate 3 1.27 0.0366 0.3896 0.0125 1.4959 72.35 457.1 297.5 23.1 35.5 13.5 Substrate 4 1.21 0.0392 0.4113 0.0132 1.4858 92.44 479.9 294.3 22.9 34.9 13.5 Mean 1.25 0.0362 0.3934 0.0126 1.4923 81.3 458.3 291.0 22.6 31.5 12.1 1 Soil water storage at field capacity in 2.6 m deep

Table 4-34: Hydraulic properties and accumulated seepage of different substrates using 60 % fine particles 2014 to 2015 in Heringen

Substrate Bd Ɵr Ɵs α n Ks Volume¹ Seepage Differences g/cm³ cm³/cm³ cm³/cm³ 1/cm cm/d mm mm % mm % Substrate 1 1.33 0.0323 0.3760 0.0180 1.4543 74.09 459.7 375.3 29.2 64.0 20.6 Substrate 2 1.35 0.0344 0.3669 0.0121 1.5004 53.98 438.0 308.1 23.9 87.9 39.9 Substrate 3 1.32 0.0358 0.3794 0.0136 1.4859 60.55 458.2 329.4 25.6 67.4 25.7 Substrate 4 1.23 0.0389 0.4068 0.0136 1.4838 85.94 478.1 306.8 23.8 47.4 18.3 Mean 1.31 0.0352 0.3820 0.0141 1.4815 65.7 459.5 332.7 25.9 73.1 28.2 1 Soil water storage at field capacity in 2.6 m deep

175

5 The Water Deficit of Evapotranspiration Covers on Potash Tailing Piles Using CropWat Carolina Bilibioa*, Oliver Hensela a Department of Agricultural and Biosystems Engineering - University of Kassel, Nordbahnhofstraße 1a, D-37213 Witzenhausen, Germany

*Corresponding author ([email protected])

5.1 Graphical abstract

5.2 Highlights

 The water deficit of evapotranspiration covers was estimated using CropWat.  The annual crop evapotranspiration was on average 642 mm.  The observed and estimated actual evapotranspiration was 461.4 and 448.3 mm/year.  A mean water deficit of 25.8 % was observed and 28.7 % was estimated over three years.  Most of the water deficit was projected for spring and summer months.

176

5.3 Abstract Evapotranspiration covers are key to minimizing water percolation in waste systems. On potash tailings piles, evapotranspiration covers are important because they may decrease the leaching of brines generated from precipitation erosion over salt tails. Considering this, the water deficit of four different potash tailings piles covering materials were evaluated using the FAO CropWat model. This study was based on a lysimeter experiment carried out in Heringen, Germany. The experiment consisted of 4 treatments and two repetitions. The treatments used different technogenic substrates made of municipal incineration wastes and coal combustion residues covered with a mixture of perennial grasses. The weather conditions were monitored at 10-min intervals using an automatic weather station and the seepage was evaluated weekly. By using the CropWat model, the effective precipitation, crop evapotranspiration, actual evapotranspiration and the water deficit were estimated over three calendar years. Further simulations using historical weather data determined the water deficit under different precipitation probabilities, 20, 50 and 80 %, and crop coefficients, varying from 0.4 to 1.0. CropWat estimated a crop evapotranspiration of 528.0 mm for 2014, 734.1 mm for 2015, and 663.6 mm for 2016. The estimated actual evapotranspiration was 452.7 mm in 2014, 435.1 mm in 2015 and 457.1 mm in 2016 (mean 448.3 mm or 68.5 % of the ground-level precipitation). The observed actual evapotranspiration was 556.8 mm in 2014, 360.7 mm in 2015 and 466.7 mm in 2016 (mean 461.4 mm or 69.3 % of the ground-level precipitation). Therefore, there was a mean estimated water deficit of 28.7 %. The observed water deficit over three calendar years was 25.8 %. Higher levels of water deficit were estimated in spring and summer months. Further simulations with historical weather means revealed the water deficit may range from 55.0 mm (8.7 %) for high precipitation levels to 157.4 mm (24.8 %) for low precipitation depths. Additionally, water deficit is associated with the crop coefficient, ranging from 0.0 mm using a constant crop coefficient of 0.4 to 105.3 mm using a constant crop coefficient of 1.0 for an average precipitation. On average the CropWat model agreed with the observed measurements and no large differences among the substrates was verified.

Keywords Effective precipitation; Drought; Perennial ryegrass; Lysimeters; crop coefficient

177

5.4 Introduction Evapotranspiration covers minimize the percolation of water through waste systems (Zhang and Sun, 2014; Barnswell and Dwyer, 2012; Zhang et al., 2009). An evapotranspiration cover consists of a water soil reservoir and a vegetated surface which transports the soil moisture back to the atmosphere (Hauser, 2009; Schnabel et al., 2012). The crops on evapotranspiration covers transpire the moisture, reduce erosion, decrease percolation and stabilize the surface of the soil or soil substitute (Rock et al., 2012; Hauser, 2009). Several factors affect crop growth, such as air temperature, soil temperature and nutrition (Gill et al., 2016). However, water is the most important abiotic stress associated with aboveground and root biomass production (Staniak and Kocoń, 2015; Vries et al., 2016; Dodd and Ryan, 2016). The moisture needed to meet the atmospheric and crop demand represents the potential crop evapotranspiration (Doorenbos and Kassam, 1979). Crop evapotranspiration is generally covered by precipitation or irrigation (Smith, 1992). When the crop water requirement is not met, there is a water deficit and the actual evapotranspiration is lower than the potential crop evapotranspiration (Doorenbos and Kassam, 1979). The effects of water stress vary with crop species and the stage of crop growth (Doorenbos and Kassam, 1979). Crop water deficit is predicted to increase due to changes in temperature and precipitation patterns caused by global climate change (Leitinger et al., 2015; Staniak and Kocoń, 2015). Researchers predict an increase of 1.4 to 5.8 oC in the global air temperature by 2100 (Staniak and Kocoń, 2015). Moreover, a decrease in the rain and snow levels is forecasted (Staniak and Kocoń, 2015; German Federal Government, 2008). In Germany, an increase of up to 3.5 oC by 2100 is expected (German Federal Government, 2008). This increase in temperature may intensify the evapotranspiration and the soil water depletion (Riediger et al., 2016). With regards to the precipitation, an increase of the winter rain and a decrease in the summer precipitation is estimated in Germany (German Federal Government, 2008). Many studies have evaluated the effects of water deficit in agricultural and bioenergy crops (Ings et al., 2013; Müller et al., 2014; López-López et al., 2018). However, in non-agricultural fields, water deficit may compromise ecological services, such as the purification of water and regulating the water cycle (Barnswell and Dwyer, 2012; Leitinger et al., 2015). Considering this, the aim of this study was to evaluate the water deficit of four evapotranspiration covers for potash tailings piles. The effective precipitation, crop evapotranspiration, actual evapotranspiration and water deficit were studied using the CropWat model (Smith, 1992). The estimated actual evapotranspiration was compared with the water balance components measured during a lysimeter experiment. This lysimeter experiment was carried out in Heringen, Germany from 2014 to 2016. The experiment considered 4 treatments and two repetitions. The treatments

178

consisted of technogenic substrates made of municipal incineration wastes and coal combustion residues. Moreover, a mixture of perennial grasses was used as a vegetation cover. The weather conditions were monitored at 10-min intervals using an automatic weather station and the seepage was evaluated weekly. Additional simulations were performed using 30 years of historical weather data, different precipitation probabilities, 20, 50, and 80 %; and crop coefficients, from 0.4-1.0. We hypothesize that the extreme weather in potash pile areas will likely increase crop evapotranspiration, actual evapotranspiration and water deficit.

5.5 Material and methods The material and methods section discusses the location and design of the lysimeter experiment. In addition, the methods to evaluate the meteorological parameters and the water balance components of the lysimeters are discussed. Later, the configuration of the CropWat model is described.

5.5.1 Experimental site and design The experiment was carried out at the Wintershal site which belongs to the integrated Werra potash plant from K+S KALI GmbH. The potash tailings pile known as “Monte Kali”, is located at 50° 53' 160'' North and 9° 59' 12'' East, 409 m altitude, in the outskirts of the Hessian city of Heringen, Germany (Figure 5-1). Heringen is located at 221 meters’ altitude and the climate is classified as Cfb (cold with no dry season, summer is temperate and there are at least four months with temperatures over 10 °C) under the Köppen-Geiger classification (Schwarz, 2016; Peel et al., 2007; Kottek et al., 2006). According to the historical records for 1961-1990, the average annual temperature in Heringen was 8.4 oC (Lamprecht, 2017) and the average annual precipitation was 684 mm (Deutsche Wetterdienst, 2017a).

179

(a)

(b)

Figure 5-1: (a) Monte Kali in the outskirts of the Hessian city of Heringen, (b) Germany Inside the confines of the experimental area, 544 m² (27.2 x 20 m), 8 three-meter-deep percolation lysimeters covering an area of 2 m², were installed. The lysimeters were filled with four different substrates: Substrate 1: 80 % household waste incineration slags; 20 % coal combustion residues; Substrate 2: 70 % household waste incineration slags; 30 % coal combustion residues; Substrate 3: 60 % household waste incineration slags; 10 % of washed sand from gravel extraction; 30 % coal combustion residues; Substrate 4: 50 % household waste incineration slags; 30 % coal combustion residues; 10 % furnace bottom ashes with particle sizes between 0.2 and 2 mm, labelled “Kesselsand”; 10% original bottom ashes with particle sizes from 0 to 6.3mm, labelled “Feinasche”. Kesselsand and Feinasche are from waste-to-energy power plants. After filling the lysimeters, a mixture of each technosol with organic compost, 0-20 mm sieve, was applied, totaling 200 tons of compost per hectare 0.3 m from the surface. In addition, different fractions of gravel were used on the bottom of the lysimeters to avoid washing-out substrates and to facilitate the drainage of percolated water. The experimental area was isolated from the stock pile with a 2 mm thick canvas. Moreover, the lysimeters were installed 1-m above a potash tailings layer with 3 % slope to facilitate the outflow of seepage water.

180

5.5.2 Meteorological data Precipitation in the experimental field was assessed with an automatic weather station, equipped with a Datalogger DLx-MET, Thies Clima (Göttingen). The weather station had a precipitation sensor with a collection area of 200 cm2 and collected precipitation every 10 minutes. Precipitation was also evaluated using 4 rain gauges installed at ground level, and 5 precipitation gauges installed at 1-meter height (Bilibio et al., 2017). Additional micrometeorological parameters were registered by the Thies-Clima weather station, such as wind speed (m/s, 3-m height), air temperature (2-m height), soil temperature (0.3-m depth), relative air humidity (2-m height) and solar radiation (2-m height). Due to technical problems with the weather station from 05.08.2016 to 18.08.2016, the weather values available for Eichhof (Bad Hersfeld) were incorporated for this interval. Eichhof (Bad Hersfeld) is located at 50o 50' 51.7'' North and 9° 41' 8.1'' East, and at circa 202 m altitude (Landesbetrieb Landwirtschaft Hessen, 2017).

5.5.3 Drainage and evapotranspiration assessment Discharge lines connected to the lysimeters drained percolated water. These lines were linked to 60 liter barrels placed in a nearby shelter. The amount of drained water was first recorded on 26 July 2013, just after the lysimeter´s saturation, and was registered weekly since then, on Thursdays, 9am-10am. The actual evapotranspiration was determined using the simplified water balance expression (Aboukhaled et al., 1982; Bilibio et al., 2011; Bethune et al., 2008), Eq. 5-1:

퐸푇푎 = 푃 − 퐷 (5-1)

Where ETa = actual evapotranspiration (mm), P = ground-level precipitation (mm), and D = drainage (mm). In water balance studies, one should consider ground-level precipitation as the incoming flux, as this precipitation represents the rain that reaches the soil (Hendriks, 2010). The weekly volumetric drainage collected from the lysimeters was considered as outgoing water flux.

5.5.4 Seeding and fertilization A seed mixture containing 65 % perennial ryegrass (Lolium perenne L.), 25 % red fescue (Festuca rubra L.) and 10 % Kentucky bluegrass (Poa pratensis L.) was used from 5 August to 26 September 2013, totaling 70 g/m2 (Schmeisky and Papke, 2013). In addition, the annual amount of fertilizer was 83 g/m2 in 2013, 193 g/m2 in 2014, 94 g/m2 in 2015, and 158 g/m2 in 2016, consisting

181

of 61 g/m2 of nitrogen, 80 g/m2 of phosphorus, 79 g/m2 of potassium and 9 g/m2 of magnesium (Schmeisky and Papke, 2013; Papke and Schmeisky, 2017).

5.5.6 CropWat configuration The crop water requirements were assessed using CropWat, version 8.0 (Smith, 1992; FAO, 2017) according to the following configuration:

5.5.6.1 Climate The monthly weather data registered at the Heringen experimental site from 2014 to 2016 were used, including the mean maximum and minimum air temperature, solar radiation, relative air humidity and wind speed. The solar radiation (W/m²) was converted to sun hours using the ET0 calculator, version 3.2 (FAO, 2014).

5.5.6.2 Rain Although the precipitation was measured in three different gauges, the CropWat model was performed using the precipitation depths registered in ground-level gauges. Additionally, the actual rainfall was corrected due to runoff loss and percolation or evaporation, which normally ranges from 10-30 % (Smith, 1992). Thus, the rainfall considered to estimate water deficits, i.e., effective rain, was 80 % of the actual rainfall measurements, as suggested by Smith (1992).

5.5.6.3 Crop The green cover was assumed to have a constant height of 0.30 m; and a root depth of 0.10 m in 2014, 0.18 m in 2015 and 0.25 m in 2016 (Papke and Schmeisky, 2017). Moreover, a constant crop coefficient of 1.0 (Allen et al., 1998; Bethune et al., 2008) was settled over four 90-day crop development stages. However, during the initial establishment of the green cover in 2014, the crop coefficient was settled to 0.40. This crop coefficient is recommended for the initial stage of alfalfa (Smith, 1992). The crop coefficient adjusts the reference evapotranspiration to the actual crop characteristics (Pereira and Alves, 2013). The maximum soil moisture depletion fraction, p, was 0.6 as suggested by ryegrass hay (Allen et al., 1998). This value represents the maximum reduction of the total available water (TAW) without causing crop stress (Allen et al., 1998). By using the soil water depletion fraction the ready water available (RAW) was determined. The total plant water availability (TAW) excludes the unavailable water due to drainage or very low matric potentials (Allen et al., 1998).

182

푇퐴푊 = 1000 (휃퐹퐶 − 휃푊푃) 푍푟 (5-2)

Where TAW is the total available water in the root zone (mm); θFC is the water content at field capacity (m³/m³); θPM is the water content at the permanent wilting point (m³/m³); Zr is the root depth (m).

푅퐴푊 = 푇퐴푊 푝 (5-3) Where: RAW is the ready available water (mm); and p is the soil water depletion fraction.

5.5.6.3 Substrates The hydraulic parameters needed as inputs in the CropWat model are presented in Table 5-1.

Table 5-1: Observed hydraulic properties of the substrates 1-4 from 0.0 to 0.64 m depth

2 Substrates Bd Ɵs α n Ks R Ɵfc Ɵpwp Δ Ɵfc-Ɵpwp g/cm³ cm³/cm³ 1/cm cm/d - cm³/cm³ cm³/cm³ mm/m Substrate 1 1.20 0.5038 0.0777 1.2103 687.4 0.9933 0.355 0.114 241.3 Substrate 2 1.21 0.4879 0.0696 1.1913 574.0 0.9910 0.361 0.129 232.3 Substrate 3 1.25 0.4736 0.0660 1.1850 463.0 0.9867 0.357 0.132 224.8 Substrate 4 1.17 0.4913 0.1025 1.1946 697.0 0.9896 0.339 0.118 221.2 Average 1.21 0.49 0.08 1.20 605.4 0.99 0.35 0.12 229.9 SD 0.03 0.01 0.02 0.01 110.1 0.00 0.01 0.01 8.9 CV 2.7 2.5 20.8 0.9 18.2 0.3 2.7 7.0 3.9

SD: standard deviation; CV: coefficient of variation The dry bulk density ranged from 1.17 g/cm³ in substrate 4 to 1.25 g/cm³ in substrate 3. These values are considered low, according to AG Boden (2005), comprising a mean of 1.21 g/cm³ and a low coefficient of variation among the substrates, 2.7 %. A coefficient of variation lower than 10 % was also found for the hydraulic parameters of the substrates, except for the inverse of the air entry value, α (alpha), Table 5-1. The volumetric water content at saturation showed a mean value of 0.49 cm³/cm³ whereas the water content at field capacity was on average 0.35 cm³/cm³ and the water content at the permanent wilting point was 0.12 cm³/cm³. The water retention curves of substrates1-4 are presented in Figure 5-2. This figure shows that the substrates presented a similar desaturation process.

183

0.6 )

3 Drainage water Plant available water Unavailable water ƟS ƟCC Ɵ Ɵ ƟPWP Ɵdry /cm CC PWP

3 0.5

S4 (cm S4 0.4 -

0.3

0.2

0.1

0.0

Volumetric water Volumetric contentS1 1 10 100 1000 10000 100000 1000000 10000000 Matric potential (hPa; log. scale) Substrate 1 Substrate 2 Substrate 3 Substrate 4

Figure 5-2: Water retention curve of the substrates 1-4 from 0.0 to 0.64 m depth With the volumetric water content at field capacity and at the permanent wilting point the plant- available water of the substrates was estimated which ranged from 221.2 mm/m in substrate 4 to 241.3 mm/m in substrate 1 (mean 229.9 mm/m; CV: 3.9%). These values are considered high, when compared with mineral soils. Clay soils have a plant-available water of 120 mm/m, silt soils 200 mm/m, and coarse sand soils 50 mm/m (Blume et al., 2016). The saturated hydraulic conductivity of the substrates ranged from 463.0 cm/d in substrate 3 to 697.0 cm/d in substrate 4 (mean 605.4 mm/m; CV: 18.2 %). These hydraulic conductivity values are considered very high (Blume et al., 2016) and are out of the CropWat model’s range. The maximum infiltration rate to be settled in CropWat was 300 mm/d (30 cm/d), which was used for all substrates.

5.5.6.4 Crop evapotranspiration

The maximum crop evapotranspiration (ETc) is estimated using the FAO Penman-Monteith reference evapotranspiration (ET0) (Allen et al., 1998). 900 ( ) ( ) (5-4) 0.408 ∆ 푅푛 − 퐺 + 훾 푇 + 273 푢2 푒푠 − 푒푎 퐸푇0 = ∆ + 훾 (1 + 0.34 푢2)

Where: ET0 is the reference evapotranspiration (mm/day), Rn is net radiation at the crop surface (MJ/m2/day), G is soil heat flux density (MJ/m2/day), T is mean daily air temperature at 2-m height

(°C), u2 is wind speed at 2-m height (m/s), es is saturation vapor pressure (kPa), ea is actual vapor pressure (kPa), es-ea is saturation vapor pressure deficit (kPa), Δ is the slope of the vapor pressure curve (kPa/°C), and is the psychrometric constant (kPa/°C).

184

After estimating the reference evapotranspiration, the crop evapotranspiration under standard conditions (ETc) is estimated using the following expression (Allen et al., 1998):

퐸푇푐 = 퐸푇표 퐾푐 (5-5) Where Kc is the crop coefficient (dimensionless).

5.5.6.5 Actual evapotranspiration The actual evapotranspiration is equal to the crop evapotranspiration up to the soil water depletion limit, p, afterwards the crop evapotranspiration will be reduced according to the expressions:

퐸푇푐 푎푑푗 = 퐾푠 퐾푐 퐸푇0 (5-6)

Where ETc adj is the adjusted crop evapotranspiration or actual evapotranspiration (mm), ET0 is the reference evapotranspiration (mm), Kc refers to the crop coefficient, Ks is the water stress coefficient (Allen et al., 1998). For water stress conditions, Ks<1, for non-water stress conditions,

Ks=1 (Allen et al., 1998) The water stress coefficient can be estimated using the expression (Allen et al., 1998): 푇퐴푊 − 푆푀퐷 푇퐴푊 − 푆푀퐷 (5-7) 퐾 = = 푠 푇퐴푊 − 푅퐴푊 (1 − 푝) 푇퐴푊

Where Ks is the transpiration reduction dependent on available soil moisture, which ranges from 0 to 1, SMD is the soil moisture depletion (mm), TAW is the total available soil moisture in the root zone (mm), p is the fraction of TAW that a crop can extract without suffering water stress (-) (Allen et al., 1998). The minimum root zone moisture depletion value is registered at field capacity and the maximum is equal to the total available water (Allen et al., 1998). The moisture content above field capacity is considered as drainage and the moisture content below the permanent wilting point is unavailable for the crops’ extraction (Allen et al., 1998). The CropWat model performs a daily water balance to determine the soil moisture depletion according to the expression:

푆푀퐷푖 = 푆푀퐷푖−1 + 퐸푇푎 − 푃푡표푡 − 퐼푟푟𝑖푔. 퐴푝푝푙. + 푅푂 + 퐷푃 (5-8)

Where SMDi is the soil moisture depletion at the day “i”; ETa is the actual evapotranspiration;

Ptotal is the total precipitation; Irr. Appl. is the irrigation depth; RO is the surface runoff; DP is the deep percolation, in mm. For the daily water balance, CropWat interpolated the monthly precipitation to 10-days interval and later distributed it in 2 applications within the 10-days interval, that is on the third and seventh day of the 10-days interval (Smith, 1992).

185

From the daily water depletion calculations, CropWat estimates the rain losses and effective rain (Smith, 1992). The total rain losses indicate the moisture that exceeded field capacity and was lost by drainage or runoff (Smith, 1992). The effective rain is the moisture used by the vegetation in evapotranspiration processes (Smith, 1992). The effective precipitation and rain losses are estimated from the water balance calculations for the entire growth period, 365 days (Smith, 1992).

5.5.7 Further simulations Additional simulations were performed considering different precipitation regimes, such as high, normal and low precipitation. A normal year corresponds to a precipitation with 50 % exceedance probability; a high precipitation level has 20 % exceedance probability; and a low precipitation level shows 80 % exceedance probability (Smith, 1992). Precipitation with an 80 % probability of exceedance is used for designing irrigation systems whereas precipitation with a 50 % probability exceedance is considered for irrigation planning (Smith, 1992). A high, normal and low precipitation level was estimated using 30 years (1987-2016) of precipitation registered in Bad Hersfeld. The Bad Hersfeld weather station, identification number 2171, is located at 50° 51' 6.84'' North, 9° 44' 16.08'' East, 272-m above sea level (Deutsche Wetterdienst, 2017a) and circa 20 km from Heringen, Werra (Deutsche Wetterdienst, 2017b). From this data, (1) the annual rainfall was tabulated; (2) the annual rain by descending magnitude was arranged; (3) a plotting position was organized using the expression (Smith, 1992): 푚 퐹 = 100 (5-9) 푎 푁 + 1 Where N is the number of records; m is the rank number and Fa is the plotting position. Afterwards the accumulated precipitation was estimated according to the different precipitation regimes using linear regression. In this sequence, the monthly precipitation was estimated according to the different degrees of probability using the following expression:

푃(80) (5-10) 푃푖(80). = 푃푖(50) 푃(50)

Where Pi(80) is the monthly low precipitation level for month i; Pi(50) is the normal average precipitation for month i. P(80) is the accumulated precipitation with an 80 % probability of exceedance. P(50) is the accumulated precipitation in a normal precipitation year (Smith, 1992). These additional simulations were performed using substrate 1 at a root depth of 30 cm. This depth generally concentrates most of the grasslands’ roots and perennial grasses (Hendrickson et al., 2013; Leitinger et al., 2015). Simulations were also performed using different crop coefficients

186

0.4-1.0. This crop coefficients represent the changes in the vegetation cover owing to integration of native species, fertilization and pest control.

5.5.8 Statistical analyses Descriptive statistics including the mean, standard deviation and coefficient of variation were used to describe the data set (Crawley, 2014; Field et al., 2012; Couto et al., 2013). A coefficient of variation lower than 10 % was considered low (Couto et al., 2013). Whereas the medium coefficient of variation was considered for variations between 10 and 20 %, a high coefficient of variation between 20 and 30 % and very high variation for CV larger than 30 % (Couto et al., 2013).

5.6 Results and discussions In the results section, the weather values registered at the lysimeter experimental site are presented, including the precipitation depths in ground-level gauges, minimum and maximum air temperature, relative air humidity, sun hours and wind speed from 2014 to 2016. The results of the CropWat simulations are also discussed. These refer to reference evapotranspiration, crop evapotranspiration, and actual evapotranspiration under rainfed conditions. The actual evapotranspiration is compared with the values observed in the lysimeter field. At last, simulations using different crop coefficients and precipitation probabilities are summarized.

5.6.1 Weather data Figure 5-3 show the monthly values of weather parameters from 2014 to 2016. The precipitation ranged from 576.5 mm in 2015 to 753.3 mm in 2014. The higher precipitation level in 2014 was due to the precipitation volume registered in July, circa 237.6 mm. This value was 295 % higher than the historical average for July, 60.1 mm (Deutsche Wetterdienst, 2017a). The additional weather parameters showed a low variation among the years, totaling a mean annual minimum air temperature of 6.4 oC; a mean maximum air temperature of 13.5 oC; a mean relative air humidity of 81.3 %; mean sun hours of 4.6 hours/day; and mean wind speed at 2-m height of 2.7 m/s. Higher sun hours, maximum and minimum air temperature were verified in summer months, whereas the relative air humidity and wind speed decreased in the summer (June, July and August), Figure 5-3.

187

30 30 C) o C) o 25 25

20 20

15 15

10 10

5 5

0 0 Maximum air temperature ( temperature Maximum air Minimum air temperature air ( temperature Minimum -5 -5 July May June July April May June April March August March January August October January October February February December November September December November Time (months) September Time (months)

10 100 9 8 80 7 6 60 5 4 40

Sun hours hours Sun(h/d) 3 2

Relative air air (%) humidty Relative 20 1 0 0 July July May June May June April April March March August August January January October October February February December December November November September Time (months) September Time (months)

6 250

5 200 2014 2015 4 150 2016 3 100 2 Precipitation (mm) Precipitation Wind speed (m/s) speed Wind 50 1

0 0 July May June July May June April April March March August August January October January October February February December November September December November September Time (months) Time (months)

Figure 5-3: Minimum and maximum air temperature, relative air humidity, sun hours, wind speed and precipitation in the lysimeter experimental site during 3 calendar years

188

5.6.2 Crop evapotranspiration and water deficit Table 5-2 and Figure 5-4 show the precipitation, effective precipitation, reference evapotranspiration, crop evapotranspiration, and water deficit for substrates 1-4 from 2014 to 2016. These parameters were equal among the substrates because the same crop and the weather parameters for substrates 1-4 were considered.

Table 5-2: Total precipitation, effective precipitation, reference evapotranspiration, crop evapotranspiration and water deficit of substrates 1-4 from 2014 to 2016 using CropWat at 10-days interval

Year P Peff ET0 ETc Water deficit mm mm mm mm mm % 2014 753.3 602.7 651.3 528.0 112.9 21.4 2015 576.5 461.2 735.9 734.1 387.6 52.8 2016 654.1 523.3 670.8 663.6 312.7 47.1 Mean 661.3 529.1 686.0 641.9 271.1 40.4 SD 88.6 70.9 44.3 104.7 142.0 16.7 CV 13.4 13.4 6.5 16.3 52.4 41.4

P: precipitation; Peff: effective precipitation; ET0: reference evapotranspiration; ETc: crop evapotranspiration; water deficit = Peff - ETc (10-days interval)

200 2014 2015 2016

150 67 83

100 80 36 86 91 41 78 56 32 50 33 4 9 Water depth (mm) (mm) Water depth 50 10 22 9 11 9 3 0 2 Jan.15 Jul. 14Jul. 15Jul. 16 Jul. Jan. 14 Jan. 16 Jan. Jun. 14Jun. 15Jun. 16Jun. Oct. 14 Oct. 15 Oct. 16 Oct. Feb. 14Feb. 15Feb. 15Sep. 16Feb. Apr. 14 Apr. 15 Apr. 16 Apr. May May 14 May 15 May 16 Dec. 14 Dec. 15 Dec. 16 Dec. Mar. 14 Mar. 15 Mar. 16 Mar. Aug. 14Aug. 14Nov. 15Aug. 15Nov. 16Aug. 16Nov. Sept. Sept. 14 Sept. 16 Time (month) Effective precipitation Water deficit Crop evapotranspiration

Figure 5-4: Effective precipitation, crop evapotranspiration and water deficit for substrates 1-4 during 2014, 2015 and 2016. Values within the columns refer to the water deficit of the respective month (mm). Water deficit = Peff. - ETc (10-days interval) The effective rain, consisting of 80 % from the total precipitation was 602.7 mm in 2014, 461.2 mm in 2015 and 523.3 mm in 2016. The FAO reference evapotranspiration accounted for 651.3

189

mm in 2014, 735.9 mm in 2015 and 670.8 mm in 2016. Whereas the potential crop evapotranspiration estimated using the FAO two step approach, ET0 x Kc (Allen et al., 1998) was 528.0 mm in 2014, 734.1 mm in 2015, and 663.6 mm in 2016. The lower crop evapotranspiration in 2014 in relation to the reference evapotranspiration is due to the initial crop coefficient, 0.4, fixed to the initial growth of the vegetation, from January to March. For the 10-days interval analyses, water deficit was determined by the difference between effective precipitation and crop evapotranspiration (Harmsen et al., 2009; Bos et al., 2009). As these parameters changed according to the years, the water deficit consequently varied from 2014 to 2016, showing a minimum value in 2014, 112.9 mm (21.4 % of the crop evapotranspiration), and a maximum value in 2015, 387.6 mm (52.8 % of the ETc). The highest water deficit was found in spring and summer months. The water deficit in spring 2014 was 22.7 mm, whereas 146.4 mm were estimated in 2015 and 99.8 mm for the spring season in 2016. Summer months presented the highest water deficits, 88.0 mm in 2014, 229.5 mm in 2015 and 163.3 mm in 2016. Figure 5-5 shows the daily crop and the actual evapotranspiration of substrates 1-4 from 2014 to 2016. From Figure 5-5, one can observe the crop and actual evapotranspiration were similar in November, December, January and February. However, during the vegetation period of the perennial grasses, approximately from April to September (Mueller et al., 2005) the actual evapotranspiration was lower than the crop evapotranspiration because the water consumption of the crops was below the the ready available water, Figure 5-6.

Substrate 1 Substrate 2

5 2014 2015 2016 5 2014 2015 2016

4 4

3 3

2 2 Water depth (mm) Waterdepth Water depth (mm) depth Water 1 1

0 0 1-Jul-14 1-Jul-15 1-Jul-16 1-Jul-14 1-Jul-15 1-Jul-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Apr-14 1-Apr-15 1-Apr-16 1-Apr-14 1-Apr-15 1-Apr-16

Time (days) Time (days)

190

Substrate 3 Substrate 4

5 2014 2015 2016 5 2014 2015 2016

4 4

3 3

2 2 Water depth (mm) Waterdepth Water depth (mm) Waterdepth 1 1

0 0 1-Jul-14 1-Jul-15 1-Jul-16 1-Jul-14 1-Jul-15 1-Jul-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Apr-14 1-Apr-15 1-Apr-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Apr-14 1-Apr-15 1-Apr-16 Time (days) Daily ETc Daily ETa

Figure 5-5: Crop evapotranspiration and actual evapotranspiration under rainfed conditions for substrates 1-4 in 2014, 2015 and 2016 Shallow rooted crops have a limited water reservoir to meet the evaporation demand of the atmosphere. This is illustrated in the CropWat simulations (Figure 5-6). In 2014, the root depth was 0.05 m during the initial stage and 0.10 m deep in mid-season (beginning of July), thus the total water available was 12.1 mm in the initial stage and 24 mm in mid-season for substrate 1. Whereas the ready water available was 7.2 mm at the initial stage and 14.5 mm in the mid-season for substrate 1. In 2015, the root depth varied from 0.10 m at the initial stage to 0.18 m in mid-season, hence the total water available for substrate 1 at the initial stage in 2015 was 24.2 mm and in mid-season it was 43.5 mm. Although the ready water available for substrate 1 was 14.5 mm at the initial stage and 26.1 mm in mid-season. In 2016, the root depth varied from 0.18 m at the beginning of the year and reached 0.25 m in mid-season. The total water available at the initial stage was 43.6 mm and in the mid-season it was 60.4 mm in substrate 1. Whereas the ready water available was 26.2 mm at the initial stage and 36.2 mm in the mid-season. Substrates 2-4 showed similar values. This analysis shows the need to establish crops on evapotranspiration covers with deep and abundant root systems to explore higher soil moisture volume. Figure 5-6 and Figure 5-7 show that under rainfed conditions water depletion reached the lowest limit of the total available water in the summer months. The lowest limit is the permanent wilting point (Ks = 0) and the highest is at field capacity (Ks = 1). Figure 5-7 presents the water stress coefficient, Ks, for substrate 1.

191

Substrate 1 Substrate 2 -20 -20 2014 2015 2016 2014 2015 2016 -10 -10 0 0 10 10 20 20 30 30 40 40 Water depth (mm) Waterdepth Water depth (mm) Waterdepth 50 50 60 60 70 70 1-Jul-14 1-Jul-15 1-Jul-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Jul-14 1-Jul-15 1-Jul-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Apr-14 1-Apr-15 1-Apr-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Apr-14 1-Apr-15 1-Apr-16 Time (days) Time (days)

Substrate 3 Substrate 4 -20 -20 2014 2015 2016 2014 2015 2016 -10 -10 θcc 0 0 10 10 20 20 30 30 40 40 Water depth (mm) Waterdepth (mm) Waterdepth 50 50 60 60 70 70 1-Jul-14 1-Jul-15 1-Jul-16 1-Jul-14 1-Jul-15 1-Jul-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Oct-14 1-Oct-15 1-Oct-16 1-Apr-14 1-Apr-15 1-Apr-16 1-Apr-14 1-Apr-15 1-Apr-16 TAW RAW ETc SMD ETa SMD

Figure 5-6: Soil moisture depletion under optimum (ETc) and rainfed (ETa) condition for substrates 1-4 during 2014, 2015 and 2016. θcc: field capacity

192

Substrate 1

2014 2015 2016 ) s

θ 1 cc

Water stress coefficient (K Water stress θpmp 0 1-Jul-14 1-Jul-15 1-Jul-16 1-Jan-14 1-Jan-15 1-Jan-16 1-Sep-14 1-Sep-15 1-Sep-16 1-Mar-14 1-Mar-15 1-Mar-16 1-Nov-14 1-Nov-15 1-Nov-16 1-May-14 1-May-15 1-May-16 Time (days) Ks ETc Ks ETa

Figure 5-7: Water stress coefficient under optimum (Ks ETc) and rainfed (Ks ETa) conditions for substrate 1 in 2014, 2015 and 2016. θcc: field capacity. θpmp: permanent wilting point The total ground-level precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit estimated using CropWat under daily water balance for the individual substrates are shown in Table 5-3 - Table 5-6. The average values for substrates 1-4 are presented in Table 5-11 and Table 5-12. The daily crop evapotranspiration was assumed to be equal to the actual evapotranspiration under optimum water availability. The water deficit was subsequently estimated by the difference between the crop evapotranspiration and the actual evapotranspiration (Doorenbos and Kassam, 1979). Under daily estimations, the crop evapotranspiration was 526.2 mm in 2014, 733.7 mm in 2015 and 666.0 in 2016 (Table 5-3 - Table 5-6). These values are very similar to the crop evapotranspiration estimated using 10-days interval (Table 5-2). The estimated actual evapotranspiration for substrates 1-4 was on average 452.7 mm in 2014, 435.1 mm in 2015 and 457.1 mm in 2016 (Table 5-11). The estimated actual evapotranspiration showed a low variation among the years (CV 2.6 %) and among the substrates (CV 0.4 %). The ratio estimated actual evapotranspiration to ground-level precipitation was 60.1 % in 2014, 74.0 % in 2015 and 69.9 % in 2016 (Table 5-11). A low variation among the years (CV 10.5 %) and among the substrates (CV 0.4 %) was found for the estimated actual evapotranspiration to ground-level precipitation ratio. With regards to the drainage, a mean value of 213.0 mm/year was found for substrates 1-4 (Table 5-11). This mean drainage represents circa 31.6 % of the ground-level precipitation over three calendar years (Table 5-11). Regarding the water deficit, a mean value of 73.5 mm was found in 2014 (14.0 % of the crop evapotranspiration), 298.6 mm in 2015 (40.7 % of the crop evapotranspiration) and 208.9 mm in 2016 (31.4 % of the crop evapotranspiration). As expected,

193

the lowest water deficit was estimated for the year with the highest amount of precipitation (2014) and the highest water deficit was projected for the year with the lowest precipitation level (2015). The mean water deficit over the three experimental years estimated using CropWat was 193.7 mm or 28.7 % of the theoretical crop evapotranspiration (Table 5-12).

Table 5-3: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 1 from 2014 to 2016 using CropWat under daily water balance

Year P ET0 ETc ETa ETa D D Water deficit mm mm mm mm % mm % mm % 2014 753.3 651.3 526.2 456.6 60.6 296.7 39.4 69.6 13.2 2015 576.5 735.9 733.7 436.6 75.7 139.9 24.3 297.1 40.5 2016 654.1 670.8 666.0 459.3 70.2 194.8 29.8 206.7 31.0 Mean 661.3 686.0 642.0 450.8 68.9 210.5 31.1 191.1 28.3 SD 88.6 44.3 105.8 12.4 7.7 79.6 7.7 114.5 13.8 CV 13.4 6.5 16.5 2.8 11.1 37.8 24.6 59.9 49.0

P: precipitation; ET0: reference evapotranspiration; ETc: crop evapotranspiration; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (daily water balance)

Table 5-4: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 2 from 2014 to 2016 using CropWat under daily water balance

Year P ET0 ETc ETa ETa D D Water deficit mm mm mm mm % mm % mm % 2014 753.3 651.3 526.2 453.6 60.2 299.7 39.8 72.6 13.8 2015 576.5 735.9 733.7 435.5 75.5 141.0 24.5 298.2 40.6 2016 654.1 670.8 666.0 457.7 70.0 196.4 30.0 208.3 31.3 Mean 661.3 686.0 642.0 448.9 68.6 212.4 31.4 193.0 28.6 SD 88.6 44.3 105.8 11.8 7.8 80.5 7.8 113.6 13.6 CV 13.4 6.5 16.5 2.6 11.3 37.9 24.7 58.8 47.7

P: precipitation; ET0: reference evapotranspiration; ETc: crop evapotranspiration; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (daily water balance)

194

Table 5-5: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 3 from 2014 to 2016 using CropWat under daily water balance

Year P ET0 ETc ETa ETa D D Water deficit mm mm mm mm % mm % mm % 2014 753.3 651.3 526.2 451.1 59.9 302.2 40.1 75.1 14.3 2015 576.5 735.9 733.7 434.0 75.3 142.5 24.7 299.7 40.8 2016 654.1 670.8 666.0 456.1 69.7 198.0 30.3 209.9 31.5 Mean 661.3 686.0 642.0 447.1 68.3 214.2 31.7 194.9 28.9 SD 88.6 44.3 105.8 11.6 7.8 81.1 7.8 113.0 13.5 CV 13.4 6.5 16.5 2.6 11.4 37.8 24.6 58.0 46.7

P: precipitation; ET0: reference evapotranspiration; ETc: crop evapotranspiration; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (daily water balance)

Table 5-6: Total precipitation, crop evapotranspiration, actual evapotranspiration, drainage and water deficit of substrate 4 from 2014 to 2016 using CropWat under daily water balance

Year P ET0 ETc ETa ETa D D Water deficit mm mm mm mm % mm % mm % 2014 753.3 651.3 526.2 449.6 59.7 303.7 40.3 76.6 14.6 2015 576.5 735.9 733.7 434.2 75.3 142.3 24.7 299.5 40.8 2016 654.1 670.8 666.0 455.2 69.6 198.9 30.4 210.8 31.7 Mean 661.3 686.0 642.0 446.3 68.2 215.0 31.8 195.6 29.0 SD 88.6 44.3 105.8 10.9 7.9 81.9 7.9 112.2 13.3 CV 13.4 6.5 16.5 2.4 11.6 38.1 24.9 57.4 45.9

P: precipitation; ET0: reference evapotranspiration; ETc: crop evapotranspiration; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (daily water balance)

The observed water balance components, i.e., precipitation, actual evapotranspiration and drainage, of substrates 1-4 are shown in Table 5-7 - Table 5-10. Moreover, the magnitude of the water deficits according to the substrates and years of evaluation is presented. A comparison between the observed and estimated actual evapotranspiration, drainage and water deficit considering the four different substrates is shown in Table 5-11 and Table 5-12.

195

Table 5-7: Observed water balance components of substrate 1 during three calendar years at the lysimeter experimental site

Year P ETa ETa D D Water deficit mm mm % mm % mm % 2014 753.3 ± 57.6 565.1 ± 28.8 75.0 ± 3.8 188.2 ± 28.8 25.0 ± 3.8 0.0 0.0 2015 576.5 ± 59.5 373.6 ± 11.1 64.8 ± 1.9 202.9 ± 11.0 35.2 ± 1.9 360.1 49.1 2016 654.1 ± 42.3 459.9 ± 7.5 70.3 ± 1.2 194.2 ± 7.5 29.7 ± 1.2 206.1 30.9 Mean 661.3 466.2 70.0 195.1 30.0 188.7 26.7 SD 88.6 95.9 5.1 7.4 5.1 180.7 24.8 CV 13.4 20.6 7.3 3.8 17.0 95.7 93.0

P: precipitation; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (annual water balance); ±: standard deviation

Table 5-8: Observed water balance components of substrate 2 during three calendar years at the lysimeter experimental site

Year P ETa ETa D D Water deficit mm mm % mm % mm % 2014 753.3 ± 57.6 565.9 ± 0.6 75.1 ± 0.1 187.5 ± 0.6 24.9 ± 0.1 0.0 0.0 2015 576.5 ± 59.5 354.7 ± 43.1 61.5 ± 7.5 221.7 ± 43.1 38.5 ± 7.5 379.0 51.7 2016 654.1 ± 42.3 455.0 ± 19.9 69.6 ± 3.0 199.1 ± 19.9 30.4 ± 3.0 211.0 31.7 Mean 661.3 458.5 68.7 202.8 31.3 196.7 27.8 SD 88.6 105.6 6.8 17.4 6.8 189.9 26.0 CV 13.4 23.0 10.0 8.6 21.9 96.6 93.8

P: precipitation; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (annual water balance); ±: standard deviation

Table 5-9: Observed water balance components of substrate 3 during three calendar years at the lysimeter experimental site

Year P ETa ETa D D Water deficit mm mm % mm % mm % 2014 753.3 ± 57.6 554.9 ± 54.6 73.7 ± 7.3 198.4 ± 54.6 26.3 ± 7.3 0.0 0.0 2015 576.5 ± 59.5 364.3 ± 2.9 63.2 ± 0.5 212.2 ± 2.9 36.8 ± 0.5 369.4 50.3 2016 654.1 ± 42.3 480.6 ± 1.6 73.5 ± 0.2 173.5 ± 1.6 26.5 ± 0.2 185.4 25.3 Mean 661.3 466.6 70.1 194.7 29.9 184.9 25.2 SD 88.6 96.1 6.0 19.6 6.0 184.7 25.2 CV 13.4 20.6 8.6 10.1 20.1 99.9 99.9

P: precipitation; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (annual water balance); ±: standard deviation

196

Table 5-10: Observed water balance components of substrate 4 during three calendar years at the lysimeter experimental site

Year P ETa ETa D D Water deficit mm mm % mm % mm % 2014 753.3 ± 57.6 541.2 ± 1.1 71.8 ± 0.1 212.1 ± 1.1 28.2 ± 0.1 0.0 0.0 2015 576.5 ± 59.5 350.2 ± 14.7 60.7 ± 2.6 226.3 ± 14.7 39.3 ± 2.6 383.5 52.3 2016 654.1 ± 42.3 471.4 ± 10.8 72.1 ± 1.6 182.6 ± 10.8 27.9 ± 1.6 194.6 29.2 Mean 661.3 454.3 68.2 207.0 31.8 192.7 27.2 SD 88.6 96.6 6.5 22.3 6.5 191.8 26.2 CV 13.4 21.3 9.5 10.8 20.4 99.5 96.4

P: precipitation; ETa: actual evapotranspiration; D: drainage; water deficit = ETc - ETa (annual water balance); ±: standard deviation

Table 5-11: Observed and estimated actual evapotranspiration (ETa) and drainage (D) of substrates 1-4 from 2014 to 2016 at the lysimeter experimental site Observed Estimated

Year ETa D ETa D mm % mm % mm % mm % 2014 556.8 ± 11.5 73.9 ± 1.5 196.6 ± 11.5 26.1 ± 1.5 452.7 ± 3.1 60.1 ± 0.4 300.6 ± 3.1 39.9 ± 0.4 2015 360.7 ± 10.4 62.6 ± 1.8 215.8 ± 10.4 37.5 ± 1.8 435.1 ± 1.2 74.0 ± 3.0 141.4 ± 1.2 24.5 ± 0.2 2016 466.7 ± 11.5 71.4 ± 1.8 187.4 ± 11.5 28.6 ± 1.8 457.1 ± 1.8 69.9 ± 0.3 197.0 ± 1.8 30.4 ± 0.4 Mean 461.4 69.3 199.9 30.7 448.3 68.0 213.0 31.6 SD 98.1 6.0 14.5 6.0 11.7 7.2 80.8 7.8 CV 21.3 8.6 7.3 19.4 2.6 10.5 37.9 24.5

ETa: actual evapotranspiration; D: drainage; ±: standard deviation (n=4 substrates)

Table 5-12: Observed and estimated water deficit of substrates 1-4 from 2014 to 2016 at the lysimeter experimental site

Water deficit Year Observed Estimated mm % mm % 2014 0.0 ± 0.0 0.0 ± 0.0 73.5 ± 3.1 14.0 ± 0.6 2015 373.0 ± 10.4 50.8 ± 1.4 298.6 ± 1.2 40.7 ± 0.2 2016 199.3 ± 11.5 29.3 ± 2.9 208.9 ± 1.8 31.4 ± 0.3 Mean 190.8 26.7 193.7 28.7 SD 186.6 25.5 113.3 13.6 CV 97.8 95.5 58.5 47.3

±: standard deviation (n=4 substrates)

197

For the observed lysimeter measurements, a low variation of the actual evapotranspiration (CV 1.3 %), drainage (CV 3.0 %) and water deficit (CV 2.7 %) was found among the substrates. The maximum actual evapotranspiration was assessed in 2014, 556.8 mm, and the lowest in 2015, 360.7 mm. The mean actual evapotranspiration was 461.4 mm/year. A medium variation for the actual evapotranspiration among the years was found (CV 21.3 %). For the drainage, on average 199.9 mm/year or circa 30.7 % of the ground-level precipitation was measured (Table 5-11). The lowest drainage was measured in 2016 (187.4 mm) and the highest in 2015 (215.8 mm). A low coefficient of variation was estimated among the years (CV 7.3 %) for the drainage. For the water deficit, a mean value of 190.8 mm/year was assessed, although the minimum water deficit was 0.0 mm in 2014 and the maximum was 373.0 mm in 2015. The ratio of water deficit to crop evapotranspiration was 0.0 % in 2014, 50.8 % in 2015 and 29.3 % in 2016 (Table 5-12). When studying the differences between CropWat and lysimeter actual evapotranspiration, the model underestimated the values under high precipitation values (2014) and overestimated under low precipitation depths (2015), Table 5-11. However, for average precipitation (2016) the model precisely estimated the actual evapotranspiration. In dry years, the differences between predicted and indirect measurements of actual evapotranspiration can be associated with additional abiotic stresses beyond the water deficit. These additional factors are, for instance, the electrical conductivity or pH of the substrates. In years with high precipitation levels, the differences between predicted and indirect measurements of actual evapotranspiration may be related with the evapotranspiration capacity of the vegetation cover. The evapotranspiration of well-watered crops may exceed the unity crop coefficient fixed in the simulation using the CropWat model. The evaluation depth can also contribute for the diferences between CropWat’s estimated actual evapotranspiration and the lysimeters assessments. The CropWat performed a daily water balance in the root zone and the lysimeters measurements were carried out at 3.0 m depth.

5.6.3 Further simulations Further simulations were performed to study the water deficit under different precipitation probabilities and crop coefficients. Figure 5-8 presents the precipitation depths on three precipitation probabilities considering a precipitation series of 30 years from Bad Hersfeld, located circa 20 km from the lysimeter experimental site. For high precipitation levels with an exceedance probability of 20 % (P20), the precipitation depth was 775 mm; whereas for a normal year (P50) the precipitation depth was circa 674.2 mm; and for a low precipitation level (P80), the precipitation depth was 573.4 mm. It is therefore possible to note the precipitation levels measured in 2014 (753.3 mm), 2015 (576.5 mm) and 2016 (654.1 mm) in Heringen, were close to a high, low and normal precipitation levels respectively. Hence, in

198

the three years of evaluation, the main water balance components of the evapotranspiration covers were evaluated under three different precipitation regimes. Even so the present simulations are valid because it considers the historical values of temperature, sun hours, wind speed and relative air humidity.

900 120 (a) (b) 800 100

80 700 60 600 40 Precipitation (mm) Precipitation 500

Dependable rain rain (mm) Dependable 20 y = -3.3594x + 842.17 400 R² = 0.9162 0 p = 0 July May June April March August January

300 October February December November 0 10 20 30 40 50 60 70 80 90 100 Time (months) September Probability of exceedance (%) P20 P50 P80

Figure 5-8: (a) Probability of exceedance of annual precipitation, points are the annual precipitations registered in Bad Hersfeld from 1987 to 2016 (30 calendar years). (b) Monthly precipitation according to different degrees of probability Figure 5-9 shows the historical average for additional weather parameters, such as minimum and maximum air temperature, sun hours, relative air humidity and wind speed in Bad Hersfeld.

30 8 100 3

25 7 80

C) 6 o 20 2 5 60 15 4 10 40

3 hours Sun(h/d) 1

5 (m/s) speed Wind 2

Air temperature ( Air temperature 20

0 1 air (%) humidityRelative 0 0 -5 0 July May June April July May June March April August January March October August February January October December February November September December November Time (months) September Time (months) Max. air temperature Min. air temperature Relative air humidity Wind speed Sun hours

Figure 5-9: (a) Minimum and maximum air temperature, sun hours; (b) relative air humidity and wind speed; in Bad Hersfeld from 1987 to 2016 (30 years)

199

Regarding the weather conditions, an annual mean value for the minimum air temperature of 4.7 oC was found, 13.7 oC for the maximum air temperature, 4.0 h/d for the sun hours, 1.9 m/s for the wind speed, and 79.4 % for the relative air humidity. These values are similar to those found in Heringen, except for the wind speed (mean 2.7 m/s in the lysimeter experiment), which can be a result of the differences in altitude. Figure 5-10 shows the effective precipitation, crop evapotranspiration and water deficit considering different exceedance probabilities of Bad Hersfeld’s precipitation. The effective precipitation was 619.9 mm for a high precipitation depth (P20), 539.3 mm for a normal year (P50) and 458.8 mm for a low precipitation depth (P80), Table 5-13. The crop evapotranspiration was equal among the years, 632.5 mm/year, because it is independent of the precipitation depths. Whereas the water deficit was 175.9 mm for P20, 221.3 mm for P50 and 267.5 mm P80, Table 5- 13. The water deficits occurred in the spring and summer, Figure 5-10. For P20, the spring water deficit was 54.0 mm, for P50 it was 68.8 mm and for P80 the spring water deficit was 83.3 mm. The summer water deficit for P20 was 112.4 mm, for a normal year 136.7 mm and for P80 the summer water deficit was 161.1 mm. Much like the observations in the lysimeter experimental field in Heringen, the water deficit increased from high annual precipitation depths to low precipitation depths.

Bad Hersfeld (S1)

200 P20 P50 P80

150

100 29 39 49 45 30 52 39 38 Water depth (mm) (mm) Water depth 59 46 45 53 10 23 50 16 1 28 22 33 3 1 5

0 July July July May June May June May June April April April March March March August August August January January January October October October February February February December December December November November November September September September Time (month) Effective precipitation Water deficit Crop evapotranspiration

Figure 5-10: Effective precipitation, crop evapotranspiration and water deficit for a green cover under different precipitation probabilities, 20, 50 and 80 %, in substrate 1

200

Table 5-13: Total precipitation, reference evapotranspiration, crop evapotranspiration and water deficit of substrate 1 for P20, P50 and P80 using CropWat at 10-days interval

Parameters Unity Probability of exceedance (%) P20 P50 P80 Precipitation mm 774.9 674.0 573.5 Effective precipitation mm 619.9 539.3 458.8 Reference evapotranspiration mm 641.0 641.3 641.0 Crop evapotranspiration mm 632.5 632.5 632.5 Water deficit mm 175.9 221.3 267.5

Figure 5-11 shows that the daily actual evapotranspiration under rainfed conditions decreased when the precipitation decreased from high (P20) to low (P80) precipitation levels. Higher differences between crop evapotranspiration and actual evapotranspiration were found from June to October (Figure 5-11, Figure 5-12). The differences between daily crop and actual evapotranspiration resulted in an estimated water deficit of 8.7 % for high precipitation levels, 16.6 % for an average precipitation depth and 24.9 % for low precipitation amounts (Table 5-14).

Table 5-14: Total precipitation, actual evapotranspiration, drainage and water deficit of substrate 1 for P20, P50 and P80 using CropWat for a rainfed field under daily water balance

Parameters Unity Probability of exceedance (%) P20 P50 P80 Precipitation mm 775.0 673.9 573.4 Actual evapotranspiration mm 577.5 527.0 474.9 Actual evapotranspiration % 74.3 78.0 82.6 Drainage mm 198.9 148.1 99.8 Water deficit % 8.7 16.6 24.9

With regards to the drainage, 198.9 mm of the seepage was estimated under high precipitation depths (P20), 148.1 mm under a normal precipitation year and 99.8 mm under low precipitation levels, Table 5-14. Totaling an actual evapotranspiration to precipitation ratio of 74.3 % for P20, 78.0 % for P50 and 82.6 % for P80 (Table 5-14). These actual evapotranspiration rates represent the water consumption of a well established crop (root depth 30 cm, crop height 30 cm) under high, normal and low precipitation levels.

201

Bad Hersfeld (S1)

5 P20 P50 P80

4

3

2

Evapotranspiration (mm) Evapotranspiration 1

0 1-Jul 1-Jul 1-Jul 1-Jan 1-Jan 1-Jan 1-Sep 1-Sep 1-Sep 1-Mar 1-Mar 1-Mar 1-Nov 1-Nov 1-Nov 1-May 1-May 1-May Time (days) Daily ETc Daily ETa Figure 5-11: Daily crop evapotranspiration and actual evapotranspiration for a green cover under different exceedance probabilities of the precipitation, 20, 50 and 80 % and substrate 1

Bad Hersfeld (S1) -20 P20 P50 P80 -10 θ 0 cc 10 20 30 40 50 Water (mm) depthWater 60 70 80 1-Jul 1-Jul 1-Jul 1-Jan 1-Jan 1-Jan 1-Sep 1-Sep 1-Sep 1-Mar 1-Mar 1-Mar 1-Nov 1-Nov 1-Nov 1-May 1-May 1-May Time (days) TAW RAW ETc SMD ETa SMD

Figure 5-12: Soil moisture depletion for a green cover under optimum (ETc SMD) and rainfed conditions (ETa SMD) considering different exceedance probabilities of the precipitation and substrate 1. θcc: field capacity Figure 5-13 shows the effective precipitation, crop evapotranspiration and water deficit for a vegetation cover under normal precipitation depths and different crop coefficients (Kc). As it is expected, it was observed that when using a constant precipitation (674.1 mm), effective precipitation (539.3 mm), reference evapotranspiration (641.0 mm), and different crop coefficients (0.4, 0.6, 0.8 and 1.0), different crop evapotranspiration depths were obtained, comprising 249.6 mm for a Kc of 0.4; 376.3 mm for a Kc of 0.6; 504.4 mm for a Kc 0.8; and 632.5 mm for a Kc of 1.0

202

(Table 5-15). This resulted in different water deficits (Peff.-ETc), 0.0 mm for a Kc of 0.4; 26.1 mm for a Kc of 0.6; 116.2 mm for a Kc of 0.8; and 221.4 mm for a Kc of 1.0. As previously observed, the water deficits were concentrated in the spring and summer months.

Bad Hersfeld (S1)

120 Kc 0.4 Kc 0.6 Kc 0.8 Kc 1.0

100

80 39 17 52 38 46 32 20

60 27 3 12 8 16 28 5 3 40 15 4 Water depth (mm) (mm) Water depth 20

0 July July July July May May May May March March March March January January January January November November November November September September September September Time (month) Effective precipitation Water deficit Crop evapotranspiration

Figure 5-13: Effective precipitation, crop evapotranspiration and water deficit for a green cover under normal precipitation depths and different crop coefficients in substrate 1

Table 5-15: Total precipitation, reference evapotranspiration, crop evapotranspiration and water deficit of substrate 1 for P50 and Kc 0.4, 0.6, 0.8, 1.0 using CropWat at 10-days interval

Parameters Unity Crop coefficient 0.4 0.6 0.8 1.0 Precipitation mm 674.1 674.1 674.1 674.1 Effective precipitation mm 539.3 539.3 539.3 539.3 Reference evapotranspiration mm 641.0 641.0 641.0 641.0 Crop evapotranspiration mm 249.6 376.3 504.4 632.5 Water deficit mm 0.0 26.1 116.2 221.4

Figure 5-14 shows the daily crop and actual evapotranspiration for a vegetation cover under normal precipitation depths and different crop coefficients. Figure 5-15 presents the water depletion in the root zone for a vegetation cover under optimum and rainfed conditions considering normal precipitation depths and different crop coefficients for substrate 1. Table 5-16 highights the water deficit according to the different crop coefficients under daily water balance. The highest water deficit was estimated using a unity crop coefficient, 1.0, consisting of 16.6 % of the crop evapotranspiration.

203

Bad Hersfeld (S1)

5 Kc 0.4 Kc 0.6 Kc 0.8 Kc 1.0

4

3

2

1 Evapotranspiration (mm) Evapotranspiration

0 1-Jul 1-Jul 1-Jul 1-Jul 1-Jan 1-Jan 1-Jan 1-Jan 1-Sep 1-Sep 1-Sep 1-Sep 1-Mar 1-Mar 1-Mar 1-Mar 1-Nov 1-Nov 1-Nov 1-Nov 1-May 1-May 1-May 1-May Time (days) Daily ETc Daily ETa

Figure 5-14: Crop evapotranspiration and actual evapotranspiration of a green cover under normal precipitation depths and different crop coefficients in substrate 1

Bad Hersfeld (S1) -20 K 0.4 K 0.6 K 0.8 K 1.0 -10 c c c c θcc 0 10 20 30 40 50 Water (mm) Water depth 60 70 80 1-Jul 1-Jul 1-Jul 1-Jul 1-Jan 1-Jan 1-Jan 1-Jan 1-Sep 1-Sep 1-Sep 1-Sep 1-Mar 1-Mar 1-Mar 1-Mar 1-Nov 1-Nov 1-Nov 1-Nov 1-May 1-May 1-May 1-May Time (days) TAW RAW ETc SMD ETa SMD

Figure 5-15: Soil moisture depletion under optimum (ETc) and rainfed (ETa) condition for normal precipitation depths and different crop coefficients in substrate 1. θcc: field capacity Table 5-16: Total precipitation, actual evapotranspiration, drainage, and water deficit of substrate 1 for P50 and Kc 0.4, 0.6, 0.8, 1.0 using CropWat for a rainfed field under daily water balance

Parameters Unity Crop coefficient 0.4 0.6 0.8 1.0 Precipitation mm 674.1 674.1 674.1 674.1 Actual evapotranspiration mm 249.5 376.1 496.7 527.0 Actual evapotranspiration % 36.9 55.7 73.6 78.0 Drainage mm 424.9 298.6 178.2 148.1 Water deficit % 0.0 0.0 1.5 16.6

204

5.7 Conclusions - The actual evapotranspiration and seepage depths using the CropWat model for the rainfed green cover on average were close to the values observed in the field. The differences between the observed and the simulated values using CropWat can be associated with additional abiotic stresses and evaluation depths. For instance, the lysimeter outflow measurements´s were performed at 3.0 m deep, whereas the seepage estimated with CropWat was assessed at the effective root depth, 0.10 m in 2014, 0.18 in 2015 and 0.25 m depth in 2016. - Reference and crop evapotranspiration are associated with the weather conditions and crop coefficients; whereas the water deficit varies according to the precipitation regimes. - Spring and summer presented the highest water deficits.

5.8 References

Aboukhaled, A., Alfaro, A., Smith, M., 1982. Lysimeters, FAO Irrigation and Drainage Paper 39, Rome. AG Boden. 2005. Bodenkundliche Kartieranleitung. (5. Aufl.). Hannover: Bundesanstalt für Geowissenschaften und Rohstoffe. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration - guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, Rome. Barnswell, K.D., Dwyer, D.F., 2012. Two-year performance by evapotranspiration covers for municipal solid waste landfills in northwest Ohio. Waste Manage. 32, 2336-2341. Doi: 10.1016/j.wasman.2012.07.014. Bethune, M.G., Selle, B., Wang, Q.J., 2008. Understanding and predicting deep percolation under surface irrigation. Water Resour. Res. 44, W12430. Doi: 10.1029/2007WR006380. Bilibio, C., Carvalho, J.A., Hensel, O., Richter, U., 2011. Effect of different levels of water deficit on rapeseed (Brassica napus L.) crop. Ciênc. Agrotec. 35, 672-684. Doi: 10.1590/S1413- 70542011000400005. Bilibio, C., Schellert, C., Retz, S., Hensel, O., Schmeisky, H., Uteau, D., Peth, S., 2017. Water balance assessment of different substrates on potash tailings piles using non-weighable lysimeters. J. Environ. Manage. 196, 633-643. Doi: 10.1016/j.jenvman.2017.01.024. Blume, H.P., Brümmer, G.W., Fleige, H., Horn, R., Kandeler, E., Kögel-Knabner, I., et al., 2016. Scheffer/Schachtschabel Soil Science. (1 ed.). Berlin: Springer. Bos, M.G., Kselik, R.A.L., Allen, R.G., Molden, D.J., 2009. Water requirements for irrigation and the environment. Dordrecht: Springer.

205

Couto, M.F., Peternelli, L.A., Barbosa, M.H.P., 2013. Classification of the coefficients of variation for sugarcane crops. Ciência Rural 43, 957-961. Doi: 10.1590/S0103-84782013000600003 Crawley, M.J., 2014. Statistics. An introduction using R. (2 ed.). Wiley. Deutsche Wetterdienst, 2017a. DWD Climate Data Center (CDC). ftp://ftp-cdc.dwd.de/pub/CDC/ (accessed 14 August 2017). Deutsche Wetterdienst, 2017b. WEtterdaten und -STatistiken Express (WESTE). http://www.dwd.de/DE/klimaumwelt/cdc/klinfo_systeme/weste/weste_node.html (accessed 01 August 2017). Dodd, I.C., Ryan, A.C., 2016. Whole-plant physiological responses to water-deficit stress. In Encyclopedia of Life Sciences. 1-9. Chichester: eLS. John Wiley & Sons, Ltd. Doorenbos, J., Kassam, A.H., 1979. Yield response to water. FAO Irrigation and Drainage Paper 33, Rome. Field, A.P., Miles, J., Field, Z., 2012. Discovering statistics using R. London: SAGE. Food and Agriculture Organization (FAO), 2017. CROPWAT. http://www.fao.org/land- water/databases-and-software/cropwat/en/ (accessed 09 September 2017).

Food and Agriculture Organization of the United Nations (FAO), 2014. ET0 calculator. Water Development and Management Unit, Information Resources. http://www.fao.org/land- water/databases-and-software/eto-calculator/en/ (accessed 17 January 2016). German Federal Government, 2008. German strategy for adaptation to climate change. http://www.bmub.bund.de/fileadmin/bmu- import/files/english/pdf/application/pdf/das_gesamt_en_bf.pdf (accessed 01 December 2017). Gill, S.S., Anjum, N.A., Gill, R., Tuteja, N., 2016. Abiotic stress signaling in plants - An overview. In Tuteja, N., Gill, S.S. (Eds.). Abiotic stress response in plants. Weinheim: Wiley-VCH GmbH & Co. Harmsen, E.W., Miller, N.L., Schlegel, N.J., Gonzalez, J.E., 2009. Seasonal climate change impacts on evapotranspiration, precipitation deficit and crop yield in Puerto Rico. Agric. Water Manage. 96, 1085-1095. Doi: 10.1016/j.agwat.2009.02.006. Hauser, V.L., 2009. Evapotranspiration covers for landfills and waste sites. Boca Raton: CRC Press. Hendrickson, J.R., Schmer, M.R., Sanderson, M.A., 2013. Water use efficiency by switchgrass compared to a native grass or a native grass alfalfa mixture. Bioenerg. Res. 6, 746-754. Doi: 10.1007/s12155-012-9290-3. Hendriks, M.R., 2010. Introduction to physical hydrology. Oxford: Oxford University Press. Ings, J., Mur, L.A.J., Robson, P.R.H., Bosch, M., 2013. Physiological and growth responses to water deficit in the bioenergy crop Miscanthus x giganteus. Front. Plant Sci. 4, 468. Doi: 10.3389/fpls.2013.00468.

206

Kottek, M., Grieser, J., Beck, C., Rudolf, B., Rubel, F., 2006. World map of the Köppen-Geiger climate classification updated. Meteorol. Z. 15, 259-263. Doi: 10.1127/0941-2948/2006/0130. Lamprecht, C., 2017. Heringen - Climate normals. Meteostat. https://www.meteostat.net/climate/heringen-hesse (accessed 28 August 2017). Landesbetrieb Landwirtschaft Hessen, 2017. Wetter. https://www.llh.hessen.de/pflanze/wetter/ (accessed 01 August 2017). Leitinger, G., Ruggenthaler, R., Hammerle, A., Lavorel, S., Schirpke, U., Clement, J., et al., 2015. Impact of droughts on water provision in managed alpine grasslands in two climatically different regions of the Alps. Ecohydrol. 8, 1600-1613. Doi: 10.1002/eco.1607. López-López, M., Espadafor, M., Testi, L., Lorite, I.J., Orgaz, F., Fereres, E., 2018. Water use of irrigated almond trees when subjected to water deficits. Agric. Water Manage. 195, 84-93. Doi: 10.1016/j.agwat.2017.10.001. Mueller, L., Behrendt, A., Schalitz, G., Schindler, U., 2005. Above ground biomass and water use efficiency of crops at shallow water tables in a temperate climate. Agric. Water Manage. 75, 117-136. Doi: 10.1016/j.agwat.2004.12.006. Müller, B.S.F., Sakamoto, T., Silveira, R.D.D., Zambussi-Carvalho, P.F., Pereira, M., Pappas, G.J.et al., 2014. Differentially expressed genes during flowering and grain filling in common bean (Phaseolus vulgaris) grown under drought stress conditions. Plant Mol. Biol. Rep. 32, 438- 451. Doi: 10.1007/s11105-013-0651-7. Papke, G., Schmeisky, H., 2017. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet – Stufe II Feldversuch Lysimeterfeld auf der Halde IV in Heringen - Endbericht Teilbericht A. Umweltsicherung Schmeisky (unveröffentlichter Bericht). Peel, M.C., Finlayson, B.L., Mcmahon, T.A., 2007. Updated world map of the Köppen-Geiger climate classification. Hydrol. Earth Syst. Sci. 11, 1633-1644. Doi: 10.5194/hess-11-1633-2007. Pereira, L.S.; Alves, I., 2013. Crop water requirements. Earth Systems and Environmental Sciences. Elsevier. Riediger, J., Breckling, B., Svoboda, N., Schröder, W., 2016. Modelling regional variability of irrigation requirements due to climate change in Northern Germany. Sci. Total Environ. 541, 329-340. Doi: 10.1016/j.scitotenv.2015.09.043. Rock, S., Myers, B., Fiedler, L., 2012. Evapotranspiration (ET) covers. Int. J. Phytorem. 14, 1-25. Doi: 10.1080/15226514.2011.609195. Schmeisky, H., Papke, G., 2013. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet – Stufe II Feldversuch Lysimeterfeld auf der Halde IV in Heringen - 1. Zwischenbericht Teilbericht A. Umweltsicherung Schmeisky (unveröffentlichter Bericht).

207

Schnabel, W.E., Munk, J., Lee, W.J., Barnes, D.L., 2012. Four-year performance evaluation of a pilot-scale evapotranspiration landfill cover in Southcentral Alaska. Cold Reg. Sci. Technol. 82, 1-7. Doi: 10.1016/j.coldregions.2012.03.009. Schwarz, T., 2016. Climate: Heringen (Werra). Climate-Data.org. https://en.climate- data.org/location/22928/ (accessed 17 January 2016). Smith, M. 1992. CROPWAT. A computer program for irrigation planning and management. FAO Irrigation and Drainage Paper 46, Rome. Staniak, M., Kocoń, A., 2015. Forage grasses under drought stress in conditions of Poland. Acta Physiol Plant 37. Doi: 10.1007/s11738-015-1864-1. Vries, F.T., Brown, C., Stevens, C.J., 2016. Grassland species root response to drought. Consequences for soil carbon and nitrogen availability. Plant Soil 409, 297-312. Doi: 10.1007/s11104-016-2964-4. Zhang, W., Sun, C., 2014. Parametric analyses of evapotranspiration landfill covers in humid regions. J. Rock Mech. Geotech. Eng. 6, 356-365. Doi: 10.1016/j.jrmge.2013.12.005. Zhang, W., Zhang, Z., Wang, K., 2009. Experimental study and simulations of infiltration in evapotranspiration landfill covers. Water Sci. Eng., 2, 96-109. Doi: 10.3882/j.issn.1674- 2370.2009.03.010.

208

General discussion

6 General discussion

The evapotranspiration and drainage of potash tailings covers were investigated throughout three studies in the present research. The first study of evapotranspiration covers’ water balance on potash tailings piles was performed in 2014 and 2015. During this initial phase, solar radiation, air and substrate temperature, wind speed and relative air humidity were evaluated. The precipitation depths at the Thies weather station, rain gauges placed at ground level and 1-m high were also assessed. By using the simplified water balance equation, ET = P – D, the actual evapotranspiration of the substrates was estimated. The estimated actual evapotranspiration from the simplified water balance equation was compared with the evapotranspiration calculated using the FAO´s two step approach (ET0 x Kc) and from Haude´s potential evapotranspiration. In this study, higher values of air temperature and solar radiation were found during the summer months (from May to October) and lower ones in winter, from November to April. Moreover, ground-level gauges registered higher precipitation than 1-m high gauges. Lower differences were measured between ground-level gauges and the Thies weather station gauges than between ground-level and 1-m high gauges. This can be due to the larger collection area of the Thies weather station rain gauge (200 cm²) than the ground-level and 1-m-high gauges (100 cm²) (Niessing, 2005). Generally, ground-level gauges represent the rain that reaches the soil, because it is not exposed to wind turbulences and higher wind speeds (Allen et al., 1998; Hoffmann et al., 2016). The actual evapotranspiration measured in this study, 454.7 mm (66.4 %) is much higher than the evaporation from potash tailings piles, circa 10 % (K+S KALI GmbH, personal communication, 2015). The mean drainage from this study was 231.7 mm (33.7 % of the ground level precipitation). The four technogenic substrates presented similar water fluxes. For the seepage, a variation of 3.6 % in 2014 and 6.4 % in 2015 was estimated. For the actual evapotranspiration, a coefficient of variation of 1.9 % in 2014 and 3.1 % in 2015 was found among the substrates. Although the water fluxes varied greatly among the years (24.1% for the seepage and 19.4 % for the actual evapotranspiration) due to the changes in weather conditions. Higher evapotranspiration rates were found in summer months, i.e., 83.5 % in 2014 and 92.6 % in 2015. In contrast, lower actual evapotranspiration was estimated in winter, i.e., 17.4 % in 2014 and 39.5 % in 2015. This demonstrates that if there is a need to collect drainage, this procedure should be concentrated from November to April. The lower evapotranspiration in winter 2014 may be due to the initial establishment of the perennial grasses, which were planted from August to September 2013 (Schmeisky and Papke, 2013). The second study simulated the water fluxes of potash tailings using Hydrus 1-D. In this study the weather parameters and water balance components of the lysimeter experiment were summarized during three water years, from 2014 to 2016. Considering the hydraulic properties of

209

General discussion the substrates control the water flow in an unsaturated profile (Hopmans, 2010; Souza et al., 2017), the hydraulic properties of substrates 1-4 were presented including the water retention curve, saturated hydraulic conductivity, particle size distribution, pore size distribution, dry bulk density, particle density, total porosity and color. Moreover, the pH and electrical conductivity were discussed. The direct and inverse simulations using Hydrus-1D were performed for 2014 and 2015. The calibrated model was validated by studying the water fluxes of a neighboring weather station, 20 km away. Additional simulations were performed considering different root depths, crop heights and rates of fine fraction. From this study the average seepage was 271.2 mm in 2014, 192.1 mm in 2015, and 231.1 mm in 2016. Moreover, the mean actual evapotranspiration was 517.1 mm in 2014, 351.7 mm in 2015 and 452.8 mm in 2016. Following the observations from the first study, higher values of evapotranspiration were found in summer than in winter. In contrast, higher seepages were found in winter than in summer (Figure 6-1). The substrates showed a mean dry bulk density of 1.21 g/cm³; a particle density of 2.56 g/cm³; and a mean total porosity of 52.8 %. These values are within the range of mineral soils. However, the mean saturated hydraulic conductivity of 605.3 cm/d is considered extremely high. Moreover, the substrates presented similar desaturation processes, with a saturation water content of 0.49 cm³/cm³, a residual water content of 0.0 cm³/cm³, an inverse value of the air entry point (alpha) 0.079 1/cm; and 1.195 for the fitting parameter of the slope of the water retention curve (n). The pH and electrical conductivity were 8.5 and 3.0 mS/cm respectively. The pH values and electrical conductivity are close to the maximum limit for mineral soils, which are 5.4-6.8 for the pH (Jones, 2012) and 2.25 to 5.6 mS/cm for electrical conductivity (Niessing, 2005; Allen et al., 1998). The direct simulation using the measured van Genuchten hydraulic properties of the substrates and weather parameters from two hydrological years showed an average seepage depth of 262.8 mm in 2014 and 121.5 mm in 2015. By optimizing the hydraulic properties of the substrates using 104 seepage measurements and values of water content a seepage depth of 269.0 mm was found in 2014 and 180.7 mm in 2015. Moreover, an actual evapotranspiration of 502.7 mm in 2014 and 429.4 mm in 2015 was predicted. The differences between the accumulated values of measured and predicted seepage was 13.6 mm (2.9 %) and for the actual evapotranspiration it was 63.3 mm (7.3 %). These seepage differences can be associated with preferential flow due to cracks and secondary pores within the substrates which are not considered in the Richards equation. The differences in actual evapotranspiration may be due to salt stress, nutrient availability and crop density.

210

General discussion

2014 2014 600 600

500 500

400 400

300 300

200 Drainage (mm) 200

100 100 Actual evapotranspiration (mm) evapotranspiration Actual 0 0 Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate Substrate Winter Summer Winter Summer

2015 2015 600 600

500 500

400 400

300 300

200 Drainage (mm) 200

100 100

Actual evapotranspiration (mm) evapotranspiration Actual 0 0 Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate Substrate Winter Summer Winter Summer

2016 2016 600 600

500 500

400 400

300 300

200 Drainage (mm) 200

100 100

Actual evapotranspiration (mm) evapotranspiration Actual 0 0 Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate Substrate Winter Summer Winter Summer Figure 6-1: Actual evapotranspiration and seepage depth during winter and summer from 2014 to 2016 for substrates 1-4. ± standard deviation

211

General discussion

The calibration increased the van Genuchten hydraulic parameters which made the water retention curve move to the left. To validate the calibrated Hydrus-1D model 27-year historical data were considered and a seepage depth of 164.8 mm and an actual evapotranspiration amount of 502.5 mm (75.3 %) were found. The simulations using different crop parameters, such as root depth and crop height revealed that any change in the crop status may alter the water fluxes. From the studies considering the different rates of fine fraction, by using bulk densities and the substrate´s texture with the Rosetta pedotransfer function, the hydraulic parameters and water fluxes using Hydrus-1D differed from the ones measured in the field. However, it was observed that an increase in the fine fractions may lead to lower seepage rates. A decrease in the seepage was also found when simulating the water fluxes according to different soil textures, such as clay, silt and sand. The next study used the relative stone content (stoniness) approach (Brakensiek et al., 1986; Bouwer and Rice, 1984) and showed that the seepage may be reduced by 7.8 % when increasing the fine fractions from 60 to 80 % in technogenic substrates. Calibrating the Hydrus-1D model with field samples showed to be the best method to make simulations with fine fractions, considering that the measurements in the laboratory are affected by the packing method, volume of the sample, particle´s size from the coarse fraction and porosity within the stones (Beckers et al., 2016; Novák, et al., 2011). The third study investigated the water deficit of the potash tailings covers. In this study, the monthly weather parameters were verified during three calendar years, considering the ground-level precipitation, maximum and minimum air temperature, relative air humidity and wind speed. Then the water deficit was determined for 10-days interval based on the reference evapotranspiration, crop evapotranspiration and effective precipitation using CropWat. In this order, the CropWat model performed a daily water balance and estimated the actual evapotranspiration of the substrates 1-4. The water deficit using daily water balance was determined by the difference between crop evapotranspiration and actual evapotranspiration. The estimated actual evapotranspiration, seepage and water deficit was later compared with the observed measurements. Additional simulations were performed using different precipitation regimes (20, 50 and 80 % exceedance probability) and crop coefficients (0.4, 0.6, 0.8 and 1.0). The magnitude of the water deficit oscillated according to the precipitation levels. In 2014, CropWat estimated a water deficit of 14.0 %, in 2015 40.7 % and in 2016 31.4 %. The mean estimated water deficit from 2014 to 2016 was 193.7 mm (28.7 %) and the observed value was 190.8 mm (26.7 %). The water deficit is also associated with the crop status. For a constant crop coefficient of 0.4, the water deficit was 0 %, whereas for a crop coefficient of 1.0, the water deficit was 16.6 %. Different aspects contribute to a decrease in crop coefficients, such as the substrate´s properties, weather, crop or pest management. Regarding the substrates, dissolved salts and pH values may interfere with the root and leaf area development (Jones, 2012). The parameters related to weather can be associated with the mechanical impacts of the wind in the 212

General discussion leaves, droughts and extreme high or low temperatures (Allen et al., 1998). Moreover, infestation of insect and natural integration of native plants and fertilization (Allen et al., 1998; Schmeisky and Hoffmann, 2000) can also affect the vegetation cover and crop coefficients. The first and the second study evaluated the observed water fluxes, seepage and actual evapotranspiration, during hydrological years. However, in the third study the water fluxes of the experiment were verified during three calendar years (from January to December). The water balance evaluated by water year and by calendar year showed little differences. The mean actual evapotranspiration measured during the calendar years was 73.9 % in 2014, 62.6 % in 2015 and 71.4 % in 2016. Whereas the actual evapotranspiration estimated using the water years was 65.6 % in 2014, 64.7 % in 2015 and 66.2 % in 2016. The lower evapotranspiration in 2014 for the water year is due to the initial establishment of the perennial grasses in November and December 2013. These months were not included in the calendar year of 2014. The weather pattern was also similar among the calendar and water years, with high temperatures and solar radiation in the summer and lower ones in the winter. Overall the use of non-weighable lysimeters was suitable for the implemented substrates owing to the particle size distribution, coarse size fraction and chemical properties of the technogenic substrates. Suction plates on the bottom, which are available for weighable lysimeters, would have been clogged due to salt leaching from substrates (Abdou and Flury, 2004). The mixture of perennial grasses is also recommended for evapotranspiration covers, because they have different growth patterns (Hauser, 2009), improve the infiltration of the substrates and reduce erosion (Wu et al., 2016; Fernando et al., 2017). Moreover, the grasses tolerate different levels of moisture stress and there is no need for reseeding (Hauser, 2009). Although Jensen et al. (2001) highlight that in the Great Basin, in the United States, the stand of perennial ryegrass decrease with time and reseeding is needed to increase plant density. Measuring the hydraulic properties of the substrates was a challenge, because the analyses were performed using a structure and methods for mineral soils with a low rate for coarse fraction. The capacity of Hydrus 1-D to adjust the data for the seepage and actual evapotranspiration rates was impressive. The use of different rain gauges provided information regarding the differences of the gauges installed during the precipitation measurements. Although these differences are not considered in official measurements, i.e., the German weather service. For the water balance assessment it is important to evaluate the water that reaches the soils. There are also a few measurement issues associated with precipitation measured at the ground level. For example, in winter the ground-level gauges covered with snow made it difficult to determine the precipitation heights during snow events. In the first study the precipitation from ground-level gauges were not adjusted, however, in the second study the precipitation of 31.12.2014, 05.02.2015, 12.02.2015, 21.01.2016 were 213

General discussion corrected assumed to be 10% higher than the 1-m high gauges. This rate was considered because similar differences between ground-levels and 1-m high gauges were found in the literature (Groh et al., 2015; Kidd and Huffman, 2011). The precipitation gauges were also weighted in the laboratory, improving the accuracy of the measurements. The studies carried out in this research contributed to verifying the evapotranspiration and drainage of potash tailings covers. The chemical properties of the seepage were evaluated monthly, however, this data will be presented in future articles. Therefore, care must be taken to extrapolate the results from this study to larger fields.

6.1 References

Abdou, H.M., Flury, M., 2004. Simulation of water flow and solute transport in free-drainage lysimeters and field soils with heterogeneous structures. Eur. J. Soil Sci. 55, 229-241. Doi: 10.1046/j.1365-2389.2004.00592.x. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration - Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, Rome. Beckers, E., Pichault, M., Pansak, W., Degré, A., Garré, S., 2016. Characterization of stony soils' hydraulic conductivity using laboratory and numerical experiments. Soil 2, 421-431. Doi: 10.5194/soil-2-421-2016. Bouwer, H., Rice, R.C., 1984. Hydraulic properties of stony vadose zones. Groundwater 22, 696- 705. Doi: 10.1111/j.1745-6584.1984.tb01438.x. Brakensiek, D.L., Rawls, W.J., Stephenson, G.R., 1986. Determining the Saturated Hydraulic Conductivity of a Soil Containing Rock Fragments. Soil Sci. Soc. Am. J. 50, 834. Doi: 10.2136/sssaj1986.03615995005000030053x. Fernando, A.L., Costa, J., Barbosa, B., Monti, A., Rettenmaier, N., 2017. Environmental impact assessment of perennial crops cultivation on marginal soils in the Mediterranean Region. Biomass Bioenergy. Doi: 10.1016/j.biombioe.2017.04.005. Groh, J., Pütz, T., Vanderborght, J.; Vereecken, H., 2015. Estimation of evapotranspiration and crop coefficient of an intensively managed grassland ecosystem with lysimeter measurements. Gumpensteiner Lysim. 16,107-112. Hauser, V.L., 2009. Evapotranspiration covers for landfills and waste sites. Boca Raton: CRC Press. Hoffmann, M., Schwartengräber, R., Wessolek, G., Peters, A., 2016. Comparison of simple rain gauge measurements with precision lysimeter data. Atmos. Res. 174-175, 120-123. Doi: 10.1016/j.atmosres.2016.01.016.

214

General discussion

Hopmans, J.W., 2010. Infiltration and unsaturated zone. In Wilderer, P.A. (Ed.). Treatise Water Sci. 103-114. Burlington: Elsevier Science. Jensen, K.B., Asay, K.H., Waldron, B.L., 2001. Dry matter production of orchardgrass and perennial ryegrass at five irrigation levels. Crop Sci. 41, 479-487. Doi: 10.2135/cropsci2001.412479x. Jones, J.B., 2012. Plant nutrition and soil fertility manual. (2 ed.). Boca Raton: CRC Press. Kidd, C., Huffman, G., 2011. Global precipitation measurement. Meteorol. Appl. 18, 334-353. Doi: 10.1002/met.284. Niessing, S., 2005. Begrünungsmaßnahmen auf der Rückstandshalde des Kaliwerkes - Sigmundshall in Bokeloh (Doctoral Dissertation). Ökologie und Umweltsicherung, Bd. 25/2005, Universität Kassel, Witzenhausen. Novák, V., Kňava, K., Šimůnek, J., 2011. Determining the influence of stones on hydraulic conductivity of saturated soils using numerical method. Geoderma 161, 177-181. Doi: 10.1016/j.geoderma.2010.12.016. Schmeisky, H., Hofmann, H., 2000. Rekultivierung von Rückstandshalden der Kaliindustrie - Untersuchungen zum Salzaustrag, zur Sukzession sowie Maßnahmen und Erkenntnisse zur Begrünung. Ökologie und Umweltsicherung, 19/2000, Universität Kassel, Witzenhausen Schmeisky, H., Papke, G., 2013. Begrünungskonzept für Kalirückstandshalden im Werra-Gebiet – Stufe II Feldversuch Lysimeterfeld auf der Halde IV in Heringen - 1. Zwischenbericht Teilbericht A. Umweltsicherung Schmeisky (unveröffentlichter Bericht). Souza, G.S., Alves, Danielle I., Dan, M.L., Lima, J.S.S., Fonseca, A.L.C.C., Araújo, J.B.S., Guimarães, L.A.O.P., 2017. Soil physico-hydraulic properties under organic conilon coffee intercropped with tree and fruit species. Pesqui. Agropecu. Bras. 52, 539-547. Doi: 10.1590/s0100-204x2017000700008. Wu, G., Yang, Z., Cui, Z., Liu, Y., Fang, N.F., Shi, Z., 2016. Mixed artificial grasslands with more roots improved mine soil infiltration capacity. J. Hydrol. 535, 54-60. Doi: 10.1016/j.jhydrol.2016.01.059

215

General conclusions

7 General conclusions

The present research aimed to evaluate the efficiency of four different potash tailings covers made of household waste incineration slags and coal combustion residues. For this, the actual evapotranspiration and drainage of technogenic substrates were assessed using non-weighable lysimeters. The experimental site was located in the outskirts of Heringen, Germany. The actual evapotranspiration of the substrates was found to be higher than the evaporation of potash tailings, circa 10 %. On average the actual evapotranspiration measured with lysimeters was 517.1 mm in 2014, 351.7 mm in 2015, and 452.8 mm in 2016 based on hydrological years. Considering the ratio actual evapotranspiration to ground-level precipitation, it was estimated an actual evapotranspiration rate of 65.6 % in 2014, 64.7 % in 2015, and 66.2 % in 2016. The summer months presented the highest evapotranspiration rates. In 2014 the summer evapotranspiration was 83.4 %, whereas 92.6 % was estimated in 2015 and 98.1 % in 2016. In contrast, the actual evapotranspiration estimated in winter months was 17.5 % in 2014, 29.2 % in 2015 and 35.8 % in 2016. The non-reference evapotranspiration (647.5 mm in 2014, 721.4 mm in 2015, 675.8 mm in 2016); crop evapotranspiration (670.8 mm in 2014, 750.9 mm in 2015, 703.1 mm in 2016); and the Haude´s potential evapotranspiration (508.9 mm in 2014, 571.7 mm in 2015, 503.2 mm in 2016) overestimated the actual evapotranspiration of the vegetation coverage, demonstrating that either water deficit or the chemical-physical properties of the substrates limited the field evapotranspiration. Seepage and actual evapotranspiration were similar among the substrates. This means that different proportions of household waste incineration slags and coal combustion residues were not consistent to influencing water fluxes. The substrates presented a mean pH value of 8.5 and 3.0 for the electrical conductivity. These values are close to the upper limit for crop growth. Further studies on particle size distribution from fine earth classifies the substrates as sandy loam, with on average 52 % of sand-size particles, 43 % of silt-size particles, and 5 % of clay-size particles. The coarse fraction comprised 42 % of the substrates’ mass. The saturated hydraulic conductivity was identified to be extremely high, with on average 605.3 cm/d up to 0.64 m depth. The calibrated Hydrus 1-D provided consistent water fluxes with low differences between observed measurements and predicted seepage and root water uptake. The calibration was better performed using the seepage and water content measurements. The validation provided similar values to the ones available in the literature. The Hydrus-1D simulations showed that fine fractions and clay-size particles may decrease seepage of the potash tailings covers. Moreover, any change in the crop status may alter the water fluxes.

216

General conclusions

The CropWater model estimated a water deficit of 73.5 mm in 2014, 298.6 mm in 2015 and 208.9 mm in 2016. Additionally, water deficit increased according to the crop growth expressed through the crop coefficients. The historical weather data showed that water deficit vary according to the weather conditions, crop coefficient and precipitation levels. Further studies should be conducted to investigate the chemical properties of the technogenic substrates made of household waste incineration slags and coal combustion residues as well as seepage properties.

217

Summary

8 Summary Potassium is one of the three main nutrients for crop production, along with nitrogen and phosphorus. Potassium is found in the potash ores of evaporites. Evaporites are mineral deposits of evaporated landlocked sea water from a former arid climate. The processing of potash ores produces millions of tons of solid and liquid wastes with a high concentration of sodium chloride. In Germany the liquids are in some cases pressed under the earth or disposed in surface waters. In contrast, the solids are often heaped above ground near the processing facilities or backfilled in mining voids. The tailings piles on the surface are exposed to precipitation erosion producing additional brines. Therefore, there is a need to implement methods to minimize the seepage of brines from potash tailings piles and reduce soil and water contamination. This can be achieved using evapotranspiration covers. Evapotranspiration covers consist of a soil reservoir and perennial plants to transport the moisture back to the atmosphere. With that in mind, a lysimeter experiment was conducted at the Wintershall potash processing facility from K+S KALI, in the outskirts of Heringen, Germany. Four different covers of potash tailings piles made of household waste incineration slags and coal combustion residues were studied. The substrates were used to fill-in 8 non-weighable lysimeters. A mixture of 65 % ryegrass (Lolium perenne L.), 25 % red fescue (Festuca rubra L.) and 10 % Kentucky bluegrass (Poa pratensis L.) was established on the substrates’ surface. The weather conditions were monitored using an automated weather station, four rain gauges placed at ground-level, and five rain gauges installed at 1-m height. The weather station registered the precipitation, air temperature, horizontal wind speed, relative air humidity, solar radiation and substrate temperature at 10 minute interval. The precipitation in ground level and 1-m-high gauges was assessed weekly. Weekly measurements also included the seepage amount, which was gathered in 60 liter barrels. The actual evapotranspiration was studied using the simplified water balance equation ETa = P - D, whereby P is precipitation and D is the drainage, both given in mm. The hydrological properties of the substrates were measured in order to calibrate Hydrus-1D and to simulate the water fluxes considering different rates of fine fractions, soil texture and crop parameters. Moreover, the water deficit of the potash tailings’ covers was verified using the CropWat model. From the observed measurements it was estimated a mean actual evapotranspiration over three hydrological years of 440.5 mm or 65.5 % for the actual evapotranspiration versus the ground-level precipitation ratio. Moreover, a mean seepage rate of 231.5 mm (34.5 %) was found between 2014 and 2016. It was verified no large differences among the substrates for seepage and actual evapotranspiration. Due to the optimal weather conditions for the crop growth, higher evapotranspiration was found in the summer (mean 363.1 mm; 91 %) than in winter time (mean 77.5 mm; 27.5 %). The calibrated Hydrus-1D model provided consistent seepage values with a variation of 2.9 % or circa 13.6 mm from the observed measurements of 218

Summary substrates 1-4 between 2014 and 2015. The calibrated model estimated an evapotranspiration rate of 75.3 % and a seepage rate of 24.7 % considering 27 years of historical weather data, which agrees with the data available in the literature. Furthermore, the simulations using Hydrus-1D suggested that an increase in fine fractions and clay-size particles may decrease the seepage of the tailings covers, as well as any change in the crop status might lead to altering the seepage and evapotranspiration values. From the CropWat study, a water deficit mainly for the spring and summer months was found. The estimated water deficit was on average 193.7 mm/year during the assessment period between 2014 and 2016. Altogether this study showed that evapotranspiration covers may decrease seepage rates of potash tailings piles. Further studies should be conducted to investigate the chemical properties of the technogenic substrates made of household waste incineration slags and coal combustion residues, as well as their specific seepage characteristics.

219

Summary

Zusammenfassung Kalium ist, neben Stickstoff und Phosphor, eines der wichtigsten Elemente für die Pflanzenernährung und -produktion. Kalium wird unter anderem in den Kaliflözen des Sedimentgesteins Evaporit gefunden. Evaporite sind Sedimentgesteine, die sich im ariden Klima zum Beispiel in Meeresbecken im Zug einer Ausfällung von Mineralien formieren. Beim Abbau von Kalisalzen fallen Millionen Tonnen an festen und flüssigen Abfällen mit einer hohen Konzentration von Natrium-Chlorid an. In Deutschland werden die Abwässer teilweise unterirdisch versenkt, oder wo möglich direkt in Oberflächengewässer eingeleitet. Die oberirdischen Feststoffe hingegen werden in der Nähe der Aufbereitungsanlage aufgehaldet oder in Bergbauhohlräumen rückverfüllt. Die Kalirückstände sind somit Niederschlagsereignissen ausgesetzt, wodurch Salzwässer entstehen. Daher müssen Technologien entwickelt und eingesetzt werden, um die Entstehung von Salzlaugen aus Kali-Rückstandshalden und die daraus resultierende Versalzung von Boden- und Wasserkörpern deutlich zu reduzieren. Dies kann zum Beispiel mit einer begrünbaren Haldenabdeckung (Evapotranspirationsabdeckung) erreicht werden. Diese besteht aus einer wasserspeichernden Bodenschicht und einem Aufwuchs aus einer Mischung verschiedener mehrjähriger Pflanzen, die die Bodenfeuchtigkeit wieder in die Atmosphäre zurücktransportieren, bevor es zur Entstehung von Sickerwasser kommen kann. Vor diesem Hintergrund wurde ein Lysimeter-Experiment am Werk Werra von der K + S KALI GmbH am Standort Wintershall in Heringen, Deutschland zwischen 2013 und 2016 durchgeführt. Es wurden vier verschiedene Abdeckvarianten, die sich aus Hausmüllverbrennungsschlacken und Kohleverbrennungsrückständen zusammensetzen, getestet. Die Substrate wurden in acht nicht wägbaren Lysimetern getestet. Eine Saatenmischung aus 65 % deutschem Weidelgras (Lolium perenne L.), 25 % gewöhnlichem Rot-Schwingel (Festuca rubra L.) und 10 % Wiesen-Rispengras (Poa pratensis L.) wurde auf der Oberfläche der Substrate etabliert. Mikrometeorologische Daten wurden automatisch von einer ThiesClima Wetterstation aufgezeichnet. Die Wetterstation registrierte den Niederschlag, die Lufttemperatur, die horizontale Windgeschwindigkeit, die relative Luftfeuchtigkeit, die Sonneneinstrahlung und die Temperatur des Substrats im Abstand von 10 Minuten. Außerdem wurde der Niederschlag mit vier Regenmessern am Boden und fünf Regenmessern in 1 m Höhe gemessen und wöchentlich bewertet. Wöchentliche Messungen umfassten auch die Sickerwassermenge, die in 60-Liter-Fässern gesammelt wurde. Die tatsächliche Evapotranspiration wurde mit der vereinfachten Wasserbilanzgleichung ETa = P - D untersucht, wobei P der Niederschlag und D die Entwässerung ist, beide angegeben in mm. Die hydraulischen Eigenschaften der in den Lysimetern eingesetzten Substrate wurden zur Kalibrierung des hydrologischen Simulationsmodells Hydrus-1D herangezogen, um mit Hilfe des genannten Modells die Wasserflüsse unter Berücksichtigung unterschiedlich feiner Fraktionen, der Textur des 220

Summary

Bodens und des Pflanzenparameters simulieren zu können. Darüber hinaus wurde das Wasserdefizit durch verschiedene Abdeckungen mit dem CropWat Modell überprüft. Aus den Messungen ergab sich eine tatsächliche Evapotranspiration von 440,5 mm oder 65,5 % des bodennah gemessenen Niederschlags über die Dauer von drei hydrologischen Messjahren. Außerdem wurde eine Gesamtsickerwassermenge von 231,5 mm (34,5 %) zwischen 2014 und 2016 dokumentiert. Aufgrund der optimalen Wetterbedingungen für das Pflanzenwachstum wurde im Sommer eine höhere Evapotranspiration (Mittelwert 363,1 mm; 91 %) als im Winter (Mittelwert 77,5 mm; 27,5 %) gemessen. Es konnten in dieser Studie keine Unterschiede zwischen den eingesetzen Substraten gefunden werden. Das kalibrierte Hydrus-1D-Modell hat Versickerungswerte mit einer Variation von 2,9 % oder ca. 13,6 mm aus den beobachteten Messungen der Substrate 1-4 zwischen 2014 und 2015 ausgegeben. Das kalibrierte Modell errechnet sich aus einer Evapotranspirationsrate von 75,3 % und eine Versickerungsrate von 24,7 % für 27 Jahre bekannter Wetterdaten des Untersuchungsgebietes, die mit der Datenlage in der Literatur übereinstimmen. Darüber hinaus konnten die Simulationen mit Hydrus-1D zeigen, dass eine Zunahme der feinen Fraktionen und Mineralpartikel der Tonfraktion sowie eine stärker ausgeprägte Vegetation (größere Wurzeltiefe und größere Blattflächen) die Versickerung auf abgedeckten Halden weiter herabsetzt. Diese Veränderung kann zu deutlich verbesserten Versickerungs- und Verdunstungswerten führen. Aus der CropWat Untersuchung ging jedoch teilweise ein Wasserdefizit vor allem für die Frühlings- und Sommermonate hervor. Die Wasserdefizite lagen im Durschnitt der Jahre 2014 – 2016 bei 193,7 mm/Jahr. Allgemein ergab sich in dieser Arbeit, dass Evapotranspirationsabdeckungen die Entstehung von salzhaltigen Wässern von Kalihalden deutlich verringern können. Darüber hinaus konnte in dieser Studie gezeigt werden, dass eine Kombination aus Hausmüllverbrennungsschlacken und Kohleverbrennungsrückständen für ein solches Abdeckverfahren grundsätzlich geeignet ist. Weitere Studien sollten jedoch durchgeführt werden, um die chemischen Eigenschaften von technogenen Substraten auf ihre spezifischen Versickerungseigenschaften hin zu untersuchen.

221

Summary

Resumo Potássio é um dos três principais elementos para produção vegetal, além do nitrogênio e o fosforo. Potássio é encontrado em agregados minerais (minérios) de evaporitos. Evaporitos são depósitos salinos, provenientes da evaporação de águas do oceano, previamente contidas nos continentes durante clima árido. O processamento de minérios de potássio produz milhões de toneladas de resíduos sólidos e líquidos com elevada concentração de cloreto de sódio. Na Alemanha, os resíduos líquidos são, em alguns casos, pressionados na subsuperfície ou são dispostos em cursos de água. Em contrapartida, os resíduos sólidos são, muitas vezes, empilhados na superfície do solo nas proximidades das instalações de processamento ou retornados aos espaços vazios da mineração subterrânea. Os resíduos sólidos na superfície são expostos à erosão hídrica produzindo drenagens salinas. Portanto, é evidente a necessidade de desenvolver métodos para reduzir a drenagem salina das pilhas de resíduo de potássio e assim reduzir a contaminação da água e do solo nas regiões onde a mineração de potássio é realizada. Isto pode ser obtido por meio de coberturas de evapotranspiração. Coberturas de evapotranspiração consistem de uma camada de solo ou material similar que funciona como um reservatório da água da chuva. Sobre a camada de solo, é estabelecido uma cobertura vegetal com culturas perenes que tem a função de transportar a umidade armazenada no solo de volta para a atmosfera. Diante do exposto, um experimento com 8 lisímetros de drenagem foi conduzido na unidade de processamento de potássio Wintershall, que pertence à empresa K+S KALI, nas proximidades da cidade de Heringen (Werra), Alemanha. O objetivo do experimento foi verificar a viabilidade do uso de coberturas de evapotranspiração para reduzir a drenagem salina de depósitos de resíduos da mineração de potássio. Quatro diferentes coberturas de evapotranspiração foram utilizadas. Os substratos incluíram o uso de diferentes proporções de resíduos municipais incinerados e resíduos da combustão de carvão. Uma mistura de diferentes gramas perenes foi utilizada na superfície dos substratos, incluindo 65 % azevém (Lolium perenne L.), 25 % festuca vermelha (Festuca rubra L.) e 10 % erva-de-febra (Poa pratensis L.). As condições ambientais foram monitoradas usando uma estação meteorológica automática, quatro pluviômetros instalados na superfície do solo, e cinco pluviômetros instalados a 1 metro de altura. A estação meteorológica registrou a precipitação, temperatura do ar, velocidade do vento horizontal, umidade relativa do ar, radiação solar e temperatura dos substratos com intervalos de 10 minutos. A precipitação nos pluviômetros localizados na superfície do solo e a um metro de altura foi avaliada semanalmente. Avaliações semanais incluíram também o volume da drenagem, que foi coletada em barris de 60 litros. A evapotranspiração atual, ETa, foi estudada utilizando a equação simplificada do balanço de água, ETa = P – D, em que P é a precipitação, D é a drenagem, ambos em mm. As propriedades físico-hídricas dos substratos foram avaliadas para calibrar o modelo matemático Hydrus-1D e para simular os fluxos de água de acordo com diferentes proporções de 222

Summary partículas finas, textura de solo, profundidade de raiz e altura de plantas. Além disso, a deficiência hídrica da cobertura vegetal foi avaliada utilizando o CropWat software. A avaliação dos resultados demonstrou uma evapotranspiração anual média de 440.5 mm ou 65.5 % da precipitação registrada nos pluviômetros instalados na superfície do solo durante três anos hidrológicos (2014 - 2016). Além disso, uma drenagem média anual de 231.5 mm (34.5 % da precipitação) foi medida para o mesmo período. Devido as condições meteorológicas apropriadas ao desenvolvimento da cultura nos meses do verão, a evapotranspiração atual foi mais elevada no verão (média 363.1 mm; 91 %) do que no inverno (média 77.5 mm; 27.5 %). Não foi verificado grandes diferenças entre os substratos para a drenagem e evapotranspiração atual. O Hydrus-1D forneceu consistente valores de drenagem com uma variação de 2.9 % ou cerca de 13.9 mm dos valores observados entre 2014 e 2015 para os substratos 1-4. O modelo estimou ainda uma evapotranspiração média de 75.3 % e uma drenagem de 24.7 % considerando 27 anos de dados meteorológicos diários, o que concorda com dados disponíveis na literatura. Além disso, simulações usando Hydrus-1D sugeriu que um aumento na proporção de terra fina (partículas menores do que 2 mm), pode contribuir para o decréscimo da drenagem de coberturas de evapotranspiração. A partir do estudo realizado com o CropWat foi estimado uma deficiência hídrica média de 193.7 mm/ano de 2014 a 2016. Uma maior deficiência hídrica foi verificada nos meses da primavera e verão. Por fim, este estudo demonstrou que coberturas de evapotranspiração podem reduzir a drenagem de pilhas de resíduos da mineração de potássio se comparados aos atuais 10 % de evaporação das pilhas de resíduos de potássio sem cobertura vegetal. Futuros estudos podem ser conduzidos para investigar as propriedades químicas dos substratos estudados, bem como as propriedades químicas da drenagem.

223

Appendix

9 Appendix

Figure 9-1: Photo documentation of the lysimeter experimental site in Heringen (Werra)

Photo: Photo: Schellert Photo: Photo: Schellert 24 October 2013 14 November 2013

Photo: Photo: Schellert Photo: Photo: Schellert 19 December 2013 30 January 2014

Photo: Photo: Schellert Photo: Photo: Schellert 06 February 2014 20 March 2014

Photo: Photo: Schellert Photo: Photo: Schellert 24 April 2014 22 May 2014

224

Appendix

Photo: Photo: Schellert Photo: Photo: Schellert 26 June 2014 31 July 2014

Photo: Photo: Bilibio Photo: Bilibio 14 August 2014 18 September 2014

Photo: Photo: Bilibio Photo: Bilibio 23 October 2014 20 November 2014

Photo: Photo: Bilibio Photo: Bilibio 11 December 2014 22 January 2015

225

Appendix

Photo: Photo: Bilibio Photo: Bilibio 19 February 2015 12 March 2015

Photo: Photo: Bilibio Photo: Bilibio 23 April 2015 21 May 2015

Photo: Photo: Bilibio Photo: Bilibio 18 June 2015 23 July 2015

Photo: Photo: Bilibio Photo: Photo: Bilibio 27 August 2015 24 September 2015

226

Appendix

Photo: Photo: Bilibio Photo: Bilibio 29 October 2015 19 November 2015

Photo: Photo: Bilibio Photo: Photo: Bilibio 10 December 2015 21 January 2016

Photo: Photo: Bilibio Photo: Bilibio 18 February 2016 17 March 2016

Photo: Photo: Bilibio Photo: Bilibio 21 April 2016 19 May 2016

227

Appendix

Photo: Photo: Bilibio Photo: Bilibio 16 June 2016 14 July 2016

Photo: Photo: Bilibio Photo: Bilibio 18 August 2016 15 September 2016

Photo: Photo: Bilibio Photo: Bilibio 20 October 2016 10 November 2016

Photo: Photo: Bilibio 15 December 2016

228

Appendix

Figure 9-2: Photo documentation of the laboratory measurements in the Agricultural and Biosystems Engineering and Soil Science laboratories of the University of Kassel

Photo: Photo: Bilibio Photo: Photo: Bilibio 21 October 2016 26 October 2016

Photo: Photo: Bilibio Photo: Bilibio 21 November 2016 27 February 2017

Photo: Photo: Bilibio Photo: Photo: Bilibio 19 June 2017 14 July 2017

229

ISSN 0931-6264 230