The effect of climate on the hydrological regime of selected Greek areas with different climate conditions

Thesis submitted in partial fulfilment of the requirements of the degree of Doctor rer. nat. of the Faculty of Environment and Natural Resources Albert-Ludwigs-University Freiburg im Breisgau, Germany

by Spyridon Paparrizos

Freiburg im Breisgau, Germany 2016

Dean: Prof. Dr. Tim Freytag Advisor: Prof. Dr. Andreas Matzarakis 2nd Advisor: Prof. Dr. Markus Weiler 3rd Advisor: Ass. Prof. Dr. Fotios Maris Oral examination date: November 7, 2016

2

To my parents Penelope and Antonios

3

Acknowledgements

By finalising the current Thesis, I would like first of all to express my gratitude to my supervisor Professor Dr. Andreas Matzarakis, especially for the continuous and unfailing supervision, as well as his priceless guidance during my life here in Freiburg. The scientific work that we performed together and our talks will be forever depicted in my Thesis, and in my memory. I would also like to thank my second supervisor, Professor Dr. Markus Weiler who has expressed kind interest about my research through our occasional talks. Special thanks need to be given to the Associate Professor Dr. Fotios Maris, my supervisor during my BSc. and MSc. studies and my mentor since my juvenile years in the university. I would be forever grateful for the time he devoted to me, as well as the resources that he provided me and I am sure that we will continue working together. I am thankful to all my colleagues of the Chair of Environmental Meteorology and first of all to PD Dr. Dirk Schindler who gave me the opportunity to work as a research assistant and assist him with some lectures. My fellow colleagues at Hebelstrasse: Dominik Fröhlich, Sven Gebhart, Ronja Vitt, Marcel Gangwisch, Shiqi Yang, Yung-Chang Chen, as well as my former colleagues Dr. Emmanuel Lubango Ndetto, Dr. Christine Ketterer and Dr. Letizia Martinelli for the substantial help that provided to me, our exciting everyday talks, lunches, and occasional feierabends; without you my life in Freiburg would have been less open- minded, less multi-cultural, and probably ennui. I want to thank also Dr. Simeon Potouridis, a colleague from my previous university and a life colleague for his interest, his support and our long talks. He urged me to come and study in Freiburg, and this was one of the best decisions I ever made. He was always there for me and I know that he will continue to be. Although the current Thesis is dedicated to my parents, I feel the need to write a few more words to emphasize their contribution. I want to thank my parents Penelope and Antonios as so my sister Emily for their substantial economic but most important: their moral support. Our parents always believed in my sister and me from the beginning of our lives, they raised us with dignity; they deprived many things in order to provide us everything. I could have never asked for a better family and I am more than thankful for that. I sincerely appreciate the help of all the teachers and professors I had since the early years of my life as each and every one of them contributed their part towards the actualization of the current Thesis, and the achievement of my goals. Finally, I want to say thanks to my friends and everyone I consider as family and have supported me all these years.

4

Summary

Assessment of present and future variations of the aspects of the hydrological cycle is essential for the research regarding climate and climate change. The current study is focused on 3 selected agricultural areas widespread in Greece, the Ardas River basin in North-eastern Greece, the Sperchios River basin in central Greece, and the Geropotamos River basin in Crete Island in South Greece that have different climate characteristics due to their location, as well as complex topography. The aim is to analyse the various aspects and procedures of the hydrological cycle, and assess its future variations, responses, and their effect on the hydrological regime of the study areas, with specific focusing on the agricultural productivity. Furthermore, the study aimed to assess the role that topography plays in the formation and spatial distribution of the climate conditions prevailing in the certain study areas. Initially, the study was focused on Potential Evapotranspiration (PET), which is one of the most critical parameters in the research on agro-ecological systems. The computational methods for the estimation of PET vary in data demands from very simple (empirically based), requiring only information based on air temperatures, to complex ones (more physically based) that require data on radiation, air humidity, etc. For this reason, 12 Potential Evapotranspiration formulae were adopted, analysed and inter-compared in terms of their sensitivity regarding their input coefficients. The aim was to compare all the methods and conclude to which empirical PET method(s) better represent the PET results in each area and thus should be adopted and used each time, and which factors influence the results in each case. Subsequently the assessment and comparison of the future variation of annual and seasonal precipitation was performed, as well as the assessment of the future response of streamflow and its impacts on the hydrological regime, in combination with other fundamental aspects of the hydrological cycle. The determination of heat requirements in the first developing phases of plants, that has been expressed as Growing Degree Days (GDD) was implemented, in order to assess the future variation and the spatial distribution of the GDD, and how these can affect the main cultivations in the study areas. Finally, the aridity and drought conditions prevailing in the study areas were estimated. Since a major task and object of the current study was to spatially interpret the various components and procedures, the current study describes a technique for downscaling climatological data in areas with limited data. In cases where the observations from the meteorological network as well as the density are insufficient to cover the entire study area and the researcher is called to operate on a regional or the mesoscale and

5 produce detailed results, the technique can be a helping hand. It constitutes a combination of statistical downscaling through multi-linear regression techniques with the dynamical downscaling through Geographic Information Systems and it was used in order to spatially interpolate with high resolution various climatological procedures in the current study. Future meteorological data were derived and analysed from a number of Regional Climate Models (RCMs) from the ENSEMBLES European project. The climate simulations were performed for the future periods 2021-2050 and 2071-2100, under the A1B and B1 emission scenarios that were developed by the Intergovernmental Panel on Climate Change. Spatial interpolation was performed using the combined dynamical and statistical downscaling technique and the Ordinary Kriging method within ArcGIS 10.2.1. Mann-Kendall statistical control method was used to investigate possible trends, while the Auto-regressive integrated moving average model (ARIMA) was used for forecasting within Mathworks version of 2014a. ArcSWAT ArcGIS extension was used to simulate the future responses of streamflow. Growing Degree Days (GDD) units was adopted to determine the present and future heat requirements of the existing cultivations. Present aridity conditions were estimated using the Aridity index (AI), while the Standardized Precipitation Index (SPI) was used to identify and assess the present and future drought conditions. According to the combined downscaling technique that was proposed in the current study, the results indicated that the current technique delivered very sufficient results as the adjusted coefficient (R2) was appeared with high values in almost every case. Areas characterized by Mediterranean type of climate with hot summers (Csa) showed the strongest presumption against null hypothesis; while areas characterized by a combination of different Mediterranean climate types (Csa and Csb) used the most coefficients in the regression equations and produced relatively good results. Areas facing continental climate conditions also delivered satisfactory results, although most of the examined independent coefficients that were used in the regression analyses were presented with medium presumption against null hypothesis. Summarizing, the described technique can be used for every type of climate in almost every terrain for the accurate representation of various climatological variables in the mesoscale. Regarding the sensitivity analysis of PET formulae, the results indicated that for the areas that face Mediterranean climate conditions, the most appropriate method for the nd estimation of PET was the temperature-based, Hamon's 2 version (PETHam2). Furthermore, the PETHam2 equation was able to estimate PET almost with the same efficiency as the average results of the 12 empirical formulae. For the Ardas River basin, the results indicated that both PETHam2 and PETHam1 can be used to estimate PET satisfactorily. Moreover, the temperature-based equations have proven to produce finer results, followed by the radiation-

6 based equations. On the other hand, PETASCE which is the most common equation, can also be applied occasionally in order to provide satisfactory results. Concerning the results of the integrated analysis and future responses of precipitation, it is expected to be critically decreased for both scenarios, future periods and study areas. Specifically, precipitation is expected to decrease by 32-40% by the end of the century in the Sperchios River basin, by almost 45% in the Ardas River basin, and more than 50% in the Geropotamos River basin. Furthermore, the decrease in precipitation for the Ardas River basin which is characterized by continental climate will be tempered, while in the Sperchios River basin the decrease will be smoother due to the influence of some minor climatic variations in the basins' springs in the highlands where milder conditions occur. Precipitation decrease in the Geropotamos River basin which is characterized by Mediterranean climate will be more vigorous. During the seasonal analysis of precipitation in the Ardas River basin, autumn season is expected to face the greatest reductions reaching a decrease of 65% by the end of the century, while for the Sperchios and the Geropotamos River basin, spring season will face the greatest reductions with almost 53% and 59%, respectively. Additionally, Mann-Kendall test indicated a strong downward trend for every study area. Respecting the results of the future assessment of streamflow, in all the examined study areas is expected to be reduced. Areas characterized by continental climate will face minor reductions by the mid-century that will become very intense by the end of it, and thus these areas will become more resistant to future changes. Specifically, the decrease in the Geropotamos River basin will range between 14-18% by the end of the century, while in the Ardas River basin it will reach up to 21.5% (A1B: 2071-2100). For the Sperchios River basin, the decrease will be stronger and by the end of the century it will reach up to 32%. This practically means that the Ardas and the Geropotamos River basins are expected to lose 1/5 of their streamflow, while the Sperchios River basin will lose more than 1/3 of its current streamflow. Areas characterized by Mediterranean climate conditions will be very vulnerable in terms of future climate change. Reduced precipitation is the main reason for decreased streamflow. Nevertheless, the high values of Actual Evapotranspiration by the end of the century due to increases in air temperature will partly equilibrate the water balance. The results regarding the Growing Degree Days indicated that for all future periods and scenarios are expected to increase. The increase in the Geropotamos River basin will be the highest reaching up to 3400 GDD units by the end of the century, followed by the Ardas and the Sperchios River basins with almost 3000 and 2250 GDD units, respectively. Moreover, the cultivation period will be shifted from April-October to April-September, which will have social, economic and environmental benefits. During the examined period of April- October which constitutes the main cultivation period in the examined study areas, the

7 results indicated that the Sperchios and Geropotamos River basins will reach up to 2700 GDD units (A1B - 2071-2100), while the Ardas River basin will reach up to 2050 GDD units. Additionally, the spatial distribution of the GDD indicated that in the upcoming years the existing cultivations can find favourable conditions and can be expanded in mountainous areas as well. On the other hand, due to the rough topography that exists in the study areas the wide expansion of the existing cultivations onto higher altitudes is unaffordable. Nevertheless, new, more profitable cultivations can be introduced which can find propitious conditions in terms of Growing Degree Days units. The results of the integrated analysis of aridity showed that the study areas are facing humid conditions, mostly due to the existence of high altitudes. The highest Aridity Index (AI) values are appeared in the Geropotamos River basin with AI = 1.09, following by the Ardas and the Sperchios River basins with 0.94 and 0.93 respectively. The various climatic conditions are responsible for differentiations in seasonal analysis regarding the aridity conditions. The study areas related to the Mediterranean climate resulted more heterogeneous conditions compared with areas affected by the continental climate as in the Geropotamos River basin the AI ranges between 0.14 (summer) to 1.72 (winter). In the Sperchios River basin the AI ranges between 0.24 (summer) to 1.80 (winter), while in the Ardas River basin the Aridity Index values range between 0.25 (summer) and 1.31 (winter). Nevertheless, the created aridity spatial maps of trend analysis presented with differentiations, especially in the mountainous areas where an extreme downward trend is appeared. Finally, the results from the integrated analysis of present and future drought conditions prevailing in the study areas indicated that for both scenarios, future periods and study areas, drought conditions are expected to be more severe in the upcoming years. The decrease of the SPI values in the Sperchios River basin is expected to be the strongest, as it is the only study area that will face a negative balance (in SPI values), regarding the drought conditions. Specifically, from the current SPI value of 0.59 it will drop to -0.79 in SPI values by the end of the century. For the Ardas and the Geropotamos River basins, a great decrease of the drought conditions will occur during the 2021-2050 period, while for 2071- 2100 period the decrease will be continued, but it will be tempered. Specifically, in the Ardas River basin the SPI values are expected to decrease from 1.14 to 0.52, while by the end of the century the will further reduce and reach to 0.32 (SPI values). In the Geropotamos River basin from the current 0.95, they values will decrease and reach by the end of the century 0.19 (in SPI values), which is a fact that indicates that the conditions in the Geropotamos River basin tend to become negative, in terms of SPI values. Nevertheless, the situation in all study areas according to the SPI classification is characterized as 'Near-normal', in terms of drought conditions.

8

Zusammenfassung

Die Bewertung der gegenwärtigen und zukünftigen Veränderungen der Komponenten des Wasserkreislaufs sind wichtig für die Untersuchung des Klimas und des Klimawandels. Die vorliegende Untersuchung konzertiert sich auf 3 ausgewählte landwirtschaftlich geprägte Gebiete, die über Griechenland verteilt sind, erhandelt sich das Ardas Einzugsgebiet im Nordosten Griechenlands, das Sperchios Einzugsgebiet in Zentral Griechenland und das Geropotamos Einzugsgebiet auf der Insel Kreta, im Süden Griechenlands, die verschiedene charakteristische Merkmale aufgrund ihrer Lage sowie ihrer komplexen Topographie aufweisen. Das Ziel ist die verschiedenen Komponenten und Prozesse des Wasserkreislaufs zu analysieren und die zukünftige Änderungen und ihre Wirkungen auf das hydrologische Regime der Untersuchungsgebiete, insbesondere bezüglich der landwirtschaftlichen Produktion zu bewerten. Darüber hinaus zielt die Untersuchung darauf, die Rolle, die die Topographie für die Bildung und die räumliche Verteilung der dominierenden klimatischen Bedingungen in der untersuchten Untersuchungsgebieten spielt, zu bewerten. Am Anfang konzentriert sich die Untersuchung auf die potentielle Evapotranspiration (PET), die eine der kritischen Bestimmungsfaktoren für die Untersuchung der agrar- ökologischen Systeme ist. Die Berechnungsmethoden für die Bewertung der PET variieren in den Anforderungen an die Daten. Von sehr einfachen empirischen Daten, welche nur Informationen über der Lufttemperatur erfordern, bis hin zu komplexeren eher physikalischen, die sich auf Strahlung und Luftfeuchtigkeit, etc. stützen. Deshalb wurden 12 PET Formeln übernommen, analysiert und in Bezug auf ihre Sensitivität bezüglich der Eingangsparameter Daten hin, vergleichen. Das Ziel war es alle Methode zu vergleichen und zu heraus zu finden welche empirische PET Methode die PET Ergebnisse am besten für den jeweiligen Bereich repräsentieren, und daher immer benutzt werden sollten. Weiterhin sollte ermittelt werden welche Faktoren die Ergebnisse im Einzelfall beeinflussen. Anschließend wurde die Bewertung und der Vergleich der zukünftigen Variation des jährlichen und saisonalen Niederschlages durchgeführt, als auch die Bewertung der zukünftigen Reaktionen des Abflusses und seine Auswirkungen auf das hydrologische Regime in Kombination mit anderen fundamentalen Komponenten des Wasserkreislaufs. Die Bestimmung der thermischen Anforderungen den ersten Entwicklungsphasen der Pflanzen, die als Growing Degree Days (GDD) zum Ausdruck gebracht wurden, wurde einbezogen, um so die zukünftigen Variationen und die räumliche Verteilung der GDD als auch deren Einfluss auf die dominierenden Landwirtschaftlichen Kulturen in den Untersuchungsgebieten zu bewerten. Schließlich wurden die Aridität und die Trockenheit in den Untersuchungsgebieten abgeschätzt. 9

Da primärer Zweck und Gegenstand der vorliegenden Untersuchung die Räumliche Interpretation verschiedener Bestandteilen und Prozesse war, beschreibt die vorliegende Untersuchung die Methodik zum Downscaling der klimatologischen Daten in Gebieten mit Datenverfügbarkeit. Die Methodik kann sehr hilfsreich sein, wenn die Aufzeichnungen des meteorologisches Netzes und die Dichte der Stationen ungenügend sind um das ganze Untersuchungsgebiet abzudecken, aber detaillierte Ergebnisse auf einer mesoskalige und regionalen Skale benötigt werden. Die Methoden setzen sich aus einer Kombination von statistischem Downscaling durch multi-lineare Regression und dynamischem Downscaling durch Geographische Informationssysteme (GIS) zusammen, um verschiedene klimatologischen Prozesse im vorliegenden Untersuchungsgebiet zu untersuchen. Klimatische Daten zukünftiger Perioden von mehreren Regionalen Klimamodellen (Regional Climate Models - RCMs) des ENSEMBLES Projekts wurden analysiert. Die Klimasimulationen wurden für die zukünftigen Perioden 2021-2050 und 2071-2100 unter Annahme der A1B und B1 Emissionsszenarien, die vom Intergovernmental Panel on Climate Change entwickelt wurden durchgeführt. Räumliche Interpolation wurde durch die Kombination dynamischer und statistischer Downscaling Methoden und die 'Ordinary Kriging' Methode in ArcGIS 10.2.1 erreicht. Der Mann-Kendall Test wurde angewendet, um mögliche Tendenzen zu untersuchen. Für die Vorhersage wurde das 'Auto-Regressive Integrated Moving Average' (ARIMA) Model in MATLAB 2014a benutzt. ArcGIS Erweiterung 'ArcSWAT' wurde angewendet, um die zukünftigen Variationen des Abflusses zu simulieren. Growing Degree Days (GDD) wurden verwendet, um die gegenwärtigen und die zukünftigen thermischen Anforderungen dominierten Landwirtschaftlichen Kulturen zu bestimmen. Gegenwärtige Aridität wurden mit Hilfe des Aridity Index (AI) bewertet. Darüber hinaus wurde der Standardized Precipitation Index (SPI) verwendet, um die gegenwärtige und zukünftige Trockenheit zu identifizieren und einzuschätzen. Die kombinierte Downscaling Methodik die in der vorliegenden Untersuchung vorgestellt wurde, führt zu hinreichend genauen Ergebnissen. Der Regressionskoeffizient (R2) weist hohe Zusammenhänge in fast allen Fällen auf. Gebiete, die als Mediterranes Klima mit heißen Sommern charakterisiert sind, weisen die stärkste Übereinstimmung der Annahme mit der Nullhypothese auf. Im Gegensatz dazu weisen Gebiete, die verschiedene Arten Mediterranen Klimas aufweisen, die größten Abweichungen zu Regressionsgleichungen auf, führen jedoch relativ guten Ergebnissen. Auch für Regionen mit Kontinentalen Klimabedingungen liefert die Methodik zufrieden stellen die Ergebnisse, obwohl die meisten der unabhängigen Regressionskoeffizienten mit einen geringeren Erwartung gegenüber der Nullhypothese verwendet wurden. Zusammenfassend kann die beschriebene Technik unabhängig von Klima und Region für die genaue Analyse von verschiedenen Variablen auf der mesoskale benutzt werden.

10

Die Ergebnisse der Sensitivitätsanalyse der PET Formel zeigen, dass die geeignetste Methode für die Abschätzung von PET in den Gebieten, die Mediterrane Klimabedingungen aufweisen, die 'Hamon's 2. Version' (PETHam2) ist. Darüber hinaus kann die PETHam2 Gleichung PET fast mit der gleichen Effizienz wie die durchschnittlichen Ergebnisse der 12 empirischen Formeln abschätzen. Bezüglich des Ardas Einzugsgebietes zeigen die

Ergebnisse, dass beide PETHam2 und PETHam1 für die hinreichende Schätzungen der PET benutzt werden können. Auf der Temperatur gestützten Gleichungen führen zu besseren Ergebnisse. Dahinter folgen die auf Strahlung basierenden Gleichungen. Andererseits kann

PETASCE, die am weitesten verbreiteteste Gleichung nur in manchen Fällen zu befriedigenden Ergebnisse führen. Die Ergebnisse der integrierten Analyse zur zukünftigen Entwicklung des Niederschlags lassen erhebliche Reduktionen für beide Szenarien gegenüber der Referenzperiode für alle Untersuchungsgebiete erwarten. Es ist zu erwarten, dass der Niederschlag bis zum Ende des Jahrhunderts in Sperchios Einzugsgebiet um 32-40%, im Ardas Einzugsgebiet um ungefähr 45% und in Geropotamos Einzugsgebiet mehr als 50% abnehmen wird. Darüber hinaus wird die Reduktion des Niederschlags in Ardas Einzugsgebiet, das ein kontinentales Klima aufweist, stark sein. Im Gegensatz dazu wird die Reduktion der Niederschlagsmenge im Sperchios Einzugsgebiet schwächer ausfallen da hier die Klimavariationen geringer im Quellgebiet schwächer ausfallen. Die Reduktion des Niederschlags in Geropotamos Einzugsgebiet das ein Mediterran Klima aufweist, wird intensiver sein. Die Ergebnisse der saisonalen Analyse des Niederschlags im Ardas Einzugsgebiet zeigen, dass in der Jahreszeit Herbst mit der größten Reduktion von bis zu 65% bis zum Ende des Jahrhunderts zu rechnen ist. Im Frühling wird sich die Niederschlagsmenge im Sperchios Einzugsgebiet und im Geropotamos Einzugsgebiet um ungefähr 53% bzw. 59% verringern. Zusätzlich zeigte der Mann-Kendall Test eine starke negative Tendenz für alle Untersuchungsgebiete. Die Ergebnisse der zukünftigen Schätzung des Abflusses zeigen eine Reduktion in allen untersuchten Untersuchungsgebieten. Regionen, die ein kontinentales Klima aufweisen, werden Mitte des Jahrhunderts mit geringen Reduktionen konfrontiert und werden daher die zukünftige Entwicklung besser verkraften. Insbesondere die Reduktion im Geropotamos Einzugsgebiet wird bis zum Ende des Jahrhunderts eine Bandbreite von 14 bis 18% aufweisen. Die Reduktion im Ardas Einzugsgebiet wird 21.5% erreichen (A1B: 2071- 2100). Für das Sperchios Einzugsgebiet wird die Reduktion stärker ausfallen und bis zum Ende des Jahrhunderts 32% erreichen. Die bedeutet, dass der Abfluss des Ardas Einzugsgebiets und des Geropotamos Einzugsgebiets bis zu 1/5 abnehmen werden. Das Sperchios Einzugsgebiet wird mehr als 1/3 seins gegenwärtigen Abflusses verlieren. Gebiete mit Mediterranen Klima sind sehr vulnerabel gegenüber den Änderungen. Der reduzierte

11

Niederschlag ist der Hauptgrund für den reduzierten Abfluss. Trotzdem werden die hohe Werte der Aktuellen Verdunstung am Ende des Jahrhunderts durch die Luftemperaturzunahme die Wasserbilanz ausgleichen. Die Ergebnisse der Analyse der Growing Degree Days (GDD) zeigen, dass für alle zukünftige Perioden und Szenarien eine Zunahme an GDD zu erwartet ist. Die Zunahme im Geropotamos Einzugsgebiet wird mit 3400 GDD am Ende des Jahrhunderts die höchste sein. Danach folgen das Ardas Einzugsgebiet und das Sperchios Einzugsgebiet mit ungefähr 3000 und 2250 GDD. Außerdem wird die Anbauperiode sich von April bis Oktober auf April bis September verschieben, was soziale, wirtschaftliche und ökologische Vorteile bringen kann. Während der untersuchte Periode von April bis Oktober des Hauptanbauperiode in den untersuchten Gebieten zeigen die Ergebnisse, dass die Sperchios und Geropotamos Einzugsgebiete bis zu 2700 GDD (A1B: 2071-2100) erreichen werden. Das Ardas Einzugsgebiet wird bis zu 2050 GDD erreichen. Zusätzlich zeigt die räumliche Verteilung der GDD, dass sich die Bedingungen für den bestehenden Anbau verbessern werden in den folgenden Jahren und sich dieser in bergige Gebiete ausdehnen kann. Aber wegen der ausgeprägten Topographie in den Untersuchungsgebieten ist die Ausdehnung des bestehenden Anbaus in großen Höhen nicht wirtschaftlich. Trotzdem kann neuer, profitablerer Anbau statt finden, da sich die Bedingungen bezüglich der GDD verbessern können. Die Ergebnisse der integrierten Analyse der Aridität haben gezeigt, dass in den Untersuchungsgebieten Feuchtigkeitszustände herrschen, die grundsätzlich auf die große Höhe zurückzuführen sind. Die höchsten Werte des Aridity Index (AI) weist das Geropotamos Einzugsgebiet mit AI=1.09 auf. Danach folgen das Ardas und das Sperchios Einzugsgebiete mit 0.94 und 0.93. Die unterschiedlichen klimatische Bedingungen sind verantwortlich für die Differenzierungen in der saisonalen Analyse der Aridität. Die Untersuchungsgebiete, die ein eher Mediterranes Klima aufweisen, zeigen stärkere Variationen als Gebiete, die vom kontinentalen Klima geprägt sind. So hat der AI im Geropotamos Einzugsgebiet eine Bandbreite von 0.14 (Sommer) bis 1.72 (Winter), im Sperchios Einzugsgebiet von 0.24 (Sommer) bis 1.80 (Winter) und im Ardas Einzugsgebiet von 0.25 (Sommer) bis 1.31 (Winter). Die Karten der Ariditätsverteilung die aus den Ergebnisse der Trendanalyse erstellt wurden, zeigen insbesondere in bergigen Regionen eine starke negative Tendenz. Schließlich zeigen die Ergebnisse der integrierten Analyse der gegenwärtigen und zukünftigen Trockenheitsbedigungen, der Untersuchungsgebiete, dass für beide Szenarien gegenüber der Referenzperiode für alle Untersuchungsgebiete, die Trockenheit in den nächsten Jahren stärker ausfällt wird. Es ist zu erwarten, dass die Abnahme der SPI Werte in dem Sperchios Einzugsgebiet am größten sein wird, da es das einzige

12

Untersuchungsgebiet ist, das mit einer negativen Bilanz konfrontiert wird. Vom gegenwärtigen SPI Wert von 0.59 wird es bis zum Ende des Jahrhunderts auf 0.79 fallen, Die Ardas und Geropotamos Einzugsgebiete werden eine große Abnahme der Trockenheitsbedigungen in den Jahren 2021-2050 erfahren. Im Ardas Einzugsgebiet ist eine Abnahme der SPI Werte von 1.14 auf 0.52 zu erwarten. Bis zum Ende des Jahrhunderts werden die SPI Werte weiter abnehmen und 0.32 erreichen. Im Geropotamos Einzugsgebiet werden die Werte ebenfalls abnehmen und von den gegenwärtigen 0.95 bis zum Endes des Jahrhunderts auf 0.19 fallen. Es lässt sich also ableiten, dass sich die Bedingungen im Geropotamos Einzugsgebiet zumindest auf die SPI Werte bezogen negativ entwickeln werden. Dennoch ist die Lage in allen Untersuchungsgebieten bezüglich der Trockenheitsbedigungen gemäß der SPI Klassifikation als 'Near-normal' charakterisiert.

13

Contents

Acknowledgements ...... 4

Summary ...... 5

Zusammenfassung ...... 9

Chapter 1 | Introduction ...... 17

1.1 Background ...... 17

1.2 Aim of the Study ...... 20

1.3 Study structure ...... 21

Chapter 2 | Literature Review ...... 23

2.1 Aquatic origin ...... 23

2.2 Global water budget ...... 23

2.3 Hydrological cycle ...... 25

2.4 Scope of Hydrometeorology ...... 27

2.5 Water resources and climate change ...... 28

2.6 Water resources, climate change and economic impacts ...... 31

2.7 Water resources, climate change and agriculture...... 32

2.8 Water resources in Greece ...... 34

Chapter 3 | Data and Methodology ...... 37

3.1 Study areas ...... 37

3.2 Climate ...... 42

3.3 Climate Data ...... 44

3.4 Data homogeneity, correlation and future data extraction ...... 45

3.5 Downscaling and spatial interpolation techniques ...... 47

3.6 Trend analysis of climatological data...... 51

3.7 Potential Evapotranspiration formulae ...... 52

3.8 Sensitivity analysis ...... 54

3.9 Aridity Index ...... 55

14

3.10 Standardized Precipitation Index ...... 56

3.11 Runoff assessment using ArcSWAT ...... 58

3.12 Growing Degree Days ...... 60

Chapter 4 | Results ...... 63

4.1 Sensitivity analysis of PET formulae ...... 63

4.2 Integrated analysis of Precipitation ...... 70

4.3 Present and future responses of runoff ...... 77

4.4 Integrated analysis of Growing Degree Days ...... 83

4.5 Integrated analysis of Aridity ...... 89

4.6 Integrated analysis of Drought ...... 94

Chapter 5 | Discussion ...... 103

5.1 Selection of the appropriate PET formula ...... 103

5.2 Assessment of precipitation responses ...... 105

5.3 Assessment of future climate change impacts on the hydrological regime ...... 107

5.4 Assessment of present and future Growing Degree Days for agriculture ...... 110

5.5 Assessment of aridity conditions ...... 112

5.6 Assessment of present and future drought conditions ...... 115

5.7 Synthesis report ...... 117

Chapter 6 | Conclusions and Outlook ...... 119

6.1 Conclusions ...... 119

6.2 Outlook ...... 121

References ...... 123

List of Figures ...... 143

List of Tables ...... 145

List of Symbols and Abbreviations ...... 147

Appendix ...... 151

15

16

Chapter 1

Introduction

1.1 Background

Water, along with air and land, are the main sources that contribute to the conservation of human's life. Water is not only a basic element for our planet's life and environment, but also a regulating factor for economic, technological, social and cultural development (Mimikou 2005). Nevertheless, according to previous reports from the Intergovernmental Panel on Climate Change (IPCC) uncertainties in climate change impacts on water resources, and droughts and floods arise for various reasons such as different scenarios of economic development, greenhouse emissions, etc. (IPCC 2007). Climate change is the greatest human challenge the world faces, as negative effects such as flooding phenomena, heat waves, forest fires, droughts, etc. have started to becoming very severe (Hillel and Rosenzweig 2002). Human activities have been affected by this change and they will continue influencing in the future (IPCC 2007). The latest IPCC reports mask the importance of the regional changes and refer that these regional to global- scale projections of future runoff remain relatively uncertain compared to other aspects of the water cycle (IPCC 2014). Although climate change is evident, changes in the aspects of the hydrological cycle and most importantly in precipitation are harder to observe and measure with the existing records, both because of the greater difficulty in sampling precipitation and also because it is expected that precipitation will have a smaller fractional change than the water vapour content of air as the climate warms (IPCC 2014; Paparrizos et al. 2016a). Some regional precipitation trends appear to be robust, but when virtually all the land area is 'filled in' using a reconstruction method (e.g. GIS-Techniques), the resulting time series of global mean land precipitation show a little change since 1900 (IPCC 2014). Additionally, regional differences are relatively high (Nastos and Zerefos 2008). Evapotranspiration (ET) constitutes another fundamental aspect of the hydrological cycle as it can significantly affect the water budget of the natural environment (i.e. approximately 62% of all precipitation falling on land is evapotranspirated) (Ampas and Baltas 2012). Evaporation, evapotranspiration, potential evapotranspiration (PET) which is defined as the amount of water that can potentially evaporate and transpirate from a

17 vegetated surface with no restrictions than the atmospheric demand (Lu et al. 2008), and crop requirements estimates are essential for the design, operation and management of irrigation projects, and are prerequisites for the optimal water resources management and especially of agricultural areas prone to water deficits (Kotsopoulos et al. 2015). In areas where water resources are vulnerable due to overexploitation from irrigation for agricultural use, it is essential to estimate crop requirements with the greatest possible precision. Moreover, in areas where irrigation is a major component of agriculture due to low precipitation, it is also of economic importance to ascertain PET as satisfactorily as possible; in fact PET is in these regions the most significant component of the hydrological budget together with precipitation (Alexandris et al. 2006). In this way, good management and planning of available water resources is attained and water requirement must be adjusted to atmospheric demand, which is related to the climatic conditions (Jabloun and Sahli 2008). Variations in precipitation and specifically decreased land precipitation in combination with increased air temperature, which enhance evapotranspiration and reduce soil moisture are important factors that have contributed to more regions experiencing drought conditions (Dai et al. 2004). Climate uncertainties results from drought and aridity phenomena are one of the major threats in contemporary water resources management (Saravi et al. 2009). Water consumption is increasing in semi-arid rural regions around the world, mainly due to developing agricultural activities. Moreover, the rapid growth of world population and uneven distribution of resources have served to escalate both the frequency and severity of natural hazards and disasters, especially in semi-arid regions (Dalezios and Bartzokas 1995). Meanwhile, by the middle of the 21st century, annual average river runoff and water availability are projected to increase as a result of climate change at high latitudes and in some wet tropical areas, and decrease over some dry regions at mid-latitudes and in the dry tropics. Many semi-arid areas (e.g. the Mediterranean basin, etc.) are particularly exposed to the impacts of climate change and are projected to suffer a decrease of water resources due to climate change (Bates et al. 2008; Paparrizos and Matzarakis 2016). Nevertheless, due to the complexity of the hydrological processes and the different basin characteristics, a big amount of input data is needed every time in order to assess the future responses of runoff; as well as complex computational techniques that they are able to spatially interpenetrate the water movement of a certain study area with high accuracy (Paparrizos et al. 2016a). The procedures and variations of the hydrological cycle as well as drought and aridity phenomena impacts concern a variety of sectors of economy, environment and society of the affected area (Chen et al. 2001; Wang et al. 2010). The socio-economic impacts of environments may arise from the interaction between natural conditions and human factors such as changes in land use, land cover and the demand and use of water. Excessive water withdrawals can exacerbate the impact of reduced water resources, and drought and aridity

18 classification, especially in areas where the water constitutes a vital coefficient for agriculture (Wang 2005; Holman 2006). In relation to the aspects of the hydrological cycle, and in order to assess and combine the future variations of the aspects of the hydrological cycle in connection with the agriculture production, the influence of the climate on plant phenology needs to be studied from the point of view of climate change, for consideration of the potential adaptation measures of plant species (Paparrizos et al. 2016d). Indeed, this interest is greater in areas where the climate conditions might force more rapid adaptation, such as in the Mediterranean area, which is expected to suffer stronger effects in terms of climate change (Giorgi and Lionello 2008; Nastos et al. 2013a). Climate variations impact society and ecosystems in a broad variety of ways and increases existing threats that have already put pressure on the environment. Changes on water resources which are intimately tied to other social issues as food supply, health, industry, transportation and ecosystem integrity can cause chain damages to property and infrastructure and to human health, which imposes heavy costs on society and the economy. Areas where various sectors rely strongly on certain temperatures and precipitation levels such as agriculture, forestry, energy and tourism are particularly affected (IPCC 2007; 2014). The Mediterranean region is one of the most diverse and sensitive ecosystems. It is situated at the southernmost tip of the northern zone of middle latitudes, which seems to be more vulnerable to changes in global warming (Nastos et al. 2013b). Greece belongs to a part of the Mediterranean region, where the accurate knowledge of the variations of the hydrological cycle is mandatory for agriculture (Dalezios et al. 2000; Tigkas et al. 2012). Hence, possible future changes in the aspects of the hydrological cycle are more critical for agriculture than the average conditions (Paparrizos et al. 2016a). The selected study areas are located in Greece and face different climate conditions. In these study areas, agriculture is the main production and economic activity. Furthermore, the three study areas are widespread in Greece and face different climate conditions which make them rather representative. A significant part of the local population is employed fully or partly in the primary sector (agriculture and livestock). Due to these facts, most of the local population uses the water from the torrential streams for irrigation. Irrigation plays an important role, due to the reduced amount of water availability from rainfalls and the farmer's inability to 'buy' water. Farming and manufacturing activities based on the agricultural activity, as well as other activities of the primary and secondary production sector create a pressing need for a study that will provide information and adapting systems regarding the future responses of the aspects of the hydrological cycle over the selected representative areas (Paparrizos et al. 2016b).

19

1.2 Aim of the Study

The overall aim of the current study concerns the combination of the hydrological and meteorological fields, in order to investigate the effect of climate and climate change in the hydrological regime of three water basins widespread in Greece. Besides, the study is focused in the mesoscale, and it will examine the importance and role of the topography and the physical characteristics that describe the natural environment of the study areas, and are interfering with the hydrometeorological procedures which take place in the certain study areas. Additionally, since the existing meteorological network in the study areas does not have the density and thus it is insufficient and unable to cover the whole area of study (Paparrizos et al. 2014), a downscaling technique is introduced and applied which can spatially interpolate with great accuracy the variables that were used in the current study, as well as other various climatological variables. Initially, a sensitivity analysis and comparison of various potential evapotranspiration formulae that were applied in these study areas was performed in order to obtain knowledge regarding to which is the most appropriate formula to use every time while performing research. Additionally, the meteorological variables that contribute in each formula were examined regarding their sensitivity and effect, each time. The main reason for the actualization of the current part of the study was the lack of available measured PET data. Hence, in order to estimate PET, empirical equations needed to be implemented. In different climates, the annual PET values significantly vary and this constitutes the selection of the appropriate empirical PET formula very important. This question is answered, through a sensitivity analysis of various PET formulae. A study regarding the sensitivity analysis of PET formulae provides useful results about which is the most appropriate method(s) to apply in regional-local study areas with different climate conditions, and which meteorological variables most strongly affect PET and thus a part of the hydrological cycle (Paparrizos et al. 2016c). Subsequently, an integrated analysis and mapping regarding the future responses of precipitation was performed, as well as a future assessment of the runoff. Afterwards, the concept of temperature was implemented and the present and future variations of the heat units, measured in growing degree days were estimated and spatially interpolated in order to achieve a better understanding of the flowering season development in the plant species, and for forecasting when flowering will occur. Finally, an integrated analysis and mapping regarding the aridity and drought conditions prevailing in the certain study area was conducted. Seasonal assessment of aridity conditions was performed only for the present period, while assessment of drought conditions was performed for the present, as well as for the future periods that were defined in the current study.

20

The results of the current study will provide the opportunity to the farmers, residents and all the stakeholders to understand the climate conditions and the variations of the hydrological cycle and the water resources prevailing in their areas and adjust their systems in order to deal with future conditions. This contributes to the sustainable development of the agricultural production for the local population related with agricultural activities or, for these who have in mind being occupied with the agriculture sector in the future. Moreover, the different climates that exist can provide an informative comparison regarding the severity of future climate change that will prevail amongst the study areas.

Summarizing, through the current thesis, the following questions will be answered:

 How can the climatological variables be spatially interpreted in areas with insufficient meteorological network and limited data?

 Which role does topography plays in the spatial distribution of the climatological variables?

 In areas with no measured Potential Evapotranspiration (PET) data, which technique should be followed for the estimation of PET?

 How will the aspects of the hydrological cycle will respond in the future in areas that face different climate conditions?

 How the future climate change will affect the operation of the hydrological regime of the study areas and which factors are mostly involved in each type of climate?

 Will the existing cultivations find favourable conditions for their expansion in the upcoming years?

1.3 Study structure

The current study consists of a total of six chapters which constitute a summary of the methods and results of the 7 peer reviewed articles. The first chapter is a general introduction to the subject matter with clear description of the background as well as the aim of the current study. The second chapter contains an extensive literature review regarding the general aspects of hydrometeorology, while the third chapter includes a description of the general characteristics of the study areas, climate and climatological data (present and future 21 assessment), statistical, GIS- and downscaling techniques, as well as spatial interpolation and description of the implemented methodology that was followed in order to complete the various tasks that were assigned in the current study. The main findings are presented and depicted in chapter four, while a comprehensive discussion of the findings is featured in chapter five. The conclusion as well as the future research outlook is provided in chapter six, while several appendices that include an extended summary of the articles are attached at the end. As a cumulative dissertation, this thesis summarises the contents of the following publications:

1. Paparrizos, S., Maris, F., Matzarakis, A., 2016: Integrated analysis of present and future responses of precipitation over selected Greek areas with different climate conditions. Atmospheric Research, 169:199-208, DOI 10.1016/j.atmosres.2015.10.004

2. Paparrizos, S., Maris, F., Matzarakis, A., 2016: Sensitivity analysis and comparison of various potential evapotranspiration formulae for selected Greek areas with different climate conditions. Theoretical and Applied Climatology, DOI 10.1007/s00704-15-1728-z

3. Paparrizos, S., Maris, F., Matzarakis, A., 2016: Integrated analysis and mapping or aridity over Greek areas with different climate conditions. Global NEST Journal, 18(1):131-145.

4. Paparrizos, S., Matzarakis. A., 2016: Assessment of future climate change impacts on the hydrological regime of selected Greek areas with different climate conditions. Hydrology Research, DOI 10.2166/nh.2016.018

5. Paparrizos, S., Matzarakis. A., 2016: Present and future assessment of Growing Degree Days over selected areas with different climate conditions. Meteorology and Atmospheric Physics, DOI 10.1007/s00703-016-0475-8

6. Paparrizos, S., Maris, F., Weiler, M., Matzarakis. A., 2016: Analysis and mapping of present and future drought conditions over Greek areas with different climate conditions. Theoretical and Applied Climatology, DOI 10.1007/s00704-016-1964-x

7. Paparrizos, S., Maris, F., Matzarakis. A., 2016: A short note for a downscaling technique for climatological data in areas with complex topography and limited data. International Journal of Engineering Research and Development (Accepted - In Press)

22

Chapter 2

Literature Review

2.1 Aquatic origin

The water origin is closely related to the processes and mechanisms that formed the earth and its solid crust. In 1949, the Soviet scientist O. Y. Schmidt in his book 'A theory of Earth's origin: Four lectures' (Schmidt 1949) argued that during the initial stages of earth forming, water molecules were solidified by cooling and pooled to form a gaseous cloud of dust, which was concentrated. The accumulated heat from the subsequent radioactive heating caused evaporation of water from the earth's core, and the water vapour as they were re-cooled, they were transformed into concentrated drops and felt into the earth as a 'hot shower'. More plausible was the argument according to which the water vapour appeared on the formatting earth's 'plasma' simultaneously with various minerals, and during the cooling period they switched into the liquid state, creating the hydrosphere. More persuasive was the theory of V.I. Vernadsky (Vernadsky 1965) according to which the amount of the water on earth has remained unchanged for a long period and the seas that flood the earth during the transgression, are merely 'splashes' of the world's ocean.

2.2 Global water budget

The amount of water that exists on planet Earth is considered stable. According to estimations from L'vovich (1979), the amount of water that it is present on Earth is estimated at 1.36 x 109 km3. About 94% of this exists as salt water of the oceans. Most of the remaining water is bound up in the polar ice caps and in glaciers (about 2%); a small part exists as sub- surface water (less than 0.01%). About 4% is found as groundwater. Only a tiny fraction of the global water (0.01%) exists as fresh water in lakes and rivers for domestic, agricultural and industrial utilization. About 0.001% of the global water exists in the atmosphere in the vapour state as clouds. Atmospheric processes have a fundamental role in the replenishment of the fresh water supply of the world through evaporation from the oceans, transport of the moisture by air currents to distant places and condensation of the vapour to form clouds followed by precipitation as rain, snow or other forms (Rakhecha and Singh

23

2009). Table 2.01 depicts the global water balance, while table 2.02 presents the distribution of the freshwater on Earth.

Table 2.01. Global Water Balance (L'vovich 1979)

Percentage (%) of Global Location Volume of Water x 106 km3 Water Oceans 1278.4 94.0 Polar Ice and Glaciers 27.2 2.0 Groundwater 54.4 4.0 Soil Water 0.082 0.006 Lakes and Rivers 0.136 0.01 Atmosphere 0.0136 0.001 Total Global Water 1360 100 Freshwater 81.736 6.0

Table 2.02. Distribution of freshwater on Earth (Kotoulas 2001)

Percentage Type of Freshwater (%) Polar Ice and Glaciers 77.23 Groundwater - depth up to 800m 9.86 Groundwater - depth from 800 - 12.35 4000m Soil moisture 0.17 Freshwater Lakes 0.35 Rivers 0.003 Hydrated mineral soil 0.001 Plants, Animals, Humans 0.003 Humidity 0.04 Total 100

It is thus observed that 77% of the freshwater is bounded in polar ice and glaciers and 22% of the freshwater is located in underground aquifers. The surface freshwater that exists in rivers and freshwater lakes has a small percentage. Nevertheless, rivers and freshwater lakes constitute the main water sources to cover the human needs.

24

The fact that rivers and freshwater lakes (surface waters) are the main sources of water seems to contrast the image that is occurred from tables 2.01 and 2.02, according to which the percentage of groundwater is higher than the surface waters. The explanation lies in the fact that the water resources cannot be reserved, but they are renewable. Hence, surface waters present a movement and thus they are renewed with much faster rate than the underground water. In other words, the great importance is not on the static image of the water storage, but rather on the dynamic image of the global water movement, which is described by the quantity of the water movements between its various forms, namely the quantities that are being transported in the hydrological cycle (Perlman et al. 2005).

2.3 Hydrological cycle

There is a continuous chain of movement and interchange of water between the oceans, the atmosphere and the land surface and below the land surface; this chain, as depicted in figure 2.01 is known as the water cycle or the hydrologic cycle.

Figure 2.01. The hydrological cycle (www.nasa-news.org) 25

The hydrologic cycle has four basic components namely:

. Evaporation and Transpiration

. Precipitation

. Runoff (Streamflow)

. Groundwater

The hydrologic cycle can be visualized as beginning with the evaporation of water from the water bodies. Evaporation occurs from oceans, lakes, streams and land surface, and transpiration from vegetation, plant leaves and forests in the form of water vapour due to the heat energy provided by solar radiation. The water vapours move upward and after condensation they form clouds at higher altitudes. While much of the clouds fall back to the oceans as precipitation, a part of the clouds is driven to the land area by winds. Through various processes, clouds cause precipitation on the land surface in different forms, such as rain, snow, hail, sleet, dew, etc. When rain falls from clouds some of it is evaporated from the falling rain drops. Part of the precipitation reaching the ground is quickly evaporated and returned to the atmosphere. The rest seeps into the soil which enriches the moisture content of the soil. Some of the soil water is used up by plants and it returned to the atmosphere by evapotranspiration. The remaining water percolates deep to become groundwater. The groundwater may come to the surface through springs and other outlets. When rainfall is heavy and the soil is saturated the water flows over the land called runoff. Surface runoff flows into streams and lakes to reach eventually back to the oceans. In this manner, the hydrologic cycle is completed (Horton 1931; Ackermann et al. 1955; Park 2001). Of the four basic components of the hydrologic cycle, the precipitation component is important to hydrometeorologists, because it represents the quantity of water received over an area for use. The quantity of water going through individual components of the hydrologic cycle can be estimated using the continuity equation (Eq. 2.01) known as water budget equation of hydrologic equation. The hydrologic equation is simply a statement of the law of conservation of mass and is given as:

(2.01)

Where: I is the inflow volume of water during a given time period

26

O is the outflow volume of water during the time period ΔS is the change in storage of the water volume during the given period

In hydrologic calculations, volumes are often expressed as average depths over a given area. Since the total water resources of the Earth is constant, the mean annual evaporation from the oceans is estimated to be 3,35,000 km3 and from inland water bodies and the land surface 65,000 km3 giving a total mean global annual value of 4,00,000 km3. This must, on average, be equal to the total quantity of water coming back to the Earth's surface in the form of precipitation (rainfall and snowfall), because the average amount of water vapour remains constant. Dividing the total volume of evaporation per year by the surface area of the Earth (4πr2 = 4 x 3.14 x 6371 x 6371 = 5.1 x 108 km2), the average annual global rainfall is approximately 800mm. The actual distribution of precipitation on the earth is non uniform. On average, the equatorial zone gets about 2000mm while the subtropical zones receive much less than the average of 800mm. In total, precipitation exceeds evaporation over the land and evaporation exceeds precipitation over the ocean surface (Rakhecha and Singh 2009).

2.4 Scope of Hydrometeorology

Hydrometeorology is an interdisciplinary field of research that combines knowledge from the atmospheric sciences and hydrology to study the transfer and exchange of water and energy between land and the lower atmosphere (Botai et al. 2015). The scope of hydrometeorology research is clearly interdisciplinary and includes analysis of the space- time properties of proxy parameters such as precipitation, evaporation, temperatures and their influence on river systems (Mora et al. 2014). The development of hydrometeorology as a science is closely linked with the increasing use of meteorology to the problems that refer to hydrology. Evaporation is commonly used to determine water losses from lakes, ponds, and reservoirs as well as in assessing the water requirements of crops. Estimates of evaporation from river basins are used for conceptual hydrological modeling (Maris et al. 2014). Deducting evaporation losses from rainfall, it is possible to estimate surface runoff. The rate at which evaporation takes place greatly depends upon the meteorological elements of temperature, radiation, humidity, vapour pressure, sunshine hours, cloud cover, and wind velocity. These meteorological elements are of great value estimating evaporation indirectly for hydrological studies where direct measurement of evaporation is not possible.

27

Precipitation data are used for the design and construction of water resources projects. Construction of dams, reservoirs across rivers for collection and storage of river runoff water to serve the needs of the people has been in vogue for many centuries. In the planning and design of these dams, long-period precipitation data are used for the estimation of water flowing in a stream. Streamflow data are vital for the assessment of the water yield of river basins and for sizing the storage capacity of dams. Finally, hydrometeorology has a wide scope in providing precipitation statistics for use in a wide range of design and construction of hydraulic structures (Rakhecha and Singh 2009).

2.5 Water resources and climate change

According to publications from the Intergovernmental Panel on Climate Change (IPCC) (Bates et al. 2008), observational records and climate projections provide abundant evidence that freshwater resources are vulnerable and have the potential to be strongly impacted by climate change, with wide-ranging consequences for human societies and ecosystems. Observed warming over several decades has been linked to changes in the large- scale hydrological cycle such as:

. Increasing atmospheric water vapour content

. Changing precipitation patterns

. Intensity and extremes

. Reduced snow cover and widespread melting of ice

. Changes in soil moisture and runoff

Precipitation changes show substantial spatial and inter-decadal variability. Over the 20th century, precipitation has mostly increased over land in high northern latitudes, while decreases have dominated from 10°S to 30°N since the 1970s. The frequency of heavy precipitation events (or proportion of total rainfall from heavy falls) has increased over most areas. Globally, the area of land classified as very dry has more than doubled since the 1970s. There have been significant decreases in water storage in mountain glaciers and Northern Hemisphere snow cover. Shifts in the amplitude and timing of runoff in glacier -and

28 snowmelt- fed rivers, and in ice-related phenomena in rivers and lakes, have been observed (IPCC 2013). Climate model simulations for the 21st century are consistent in projecting precipitation increases in high latitudes and parts of the tropics, and decreases in some subtropical and lower mid-latitude regions. Outside these areas, the sign and magnitude of projected changes varies between models, leading to substantial uncertainty in precipitation projections. Thus projections of future precipitation changes are more robust for some regions than for others. Projections become less consistent between models as spatial scales decrease (Yin 2005). Increased precipitation intensity and variability are projected to increase the risks of flooding and drought in many areas. The frequency of heavy precipitation events (or proportion of total rainfall from heavy falls) will be very likely to increase over most areas during the 21st century, with consequences for the risk of rain-generated floods. At the same time, the proportion of land surface in extreme drought at any time is projected to increase, in addition to a tendency for drying in continental interiors during summer, especially in the sub- tropics, low and mid-latitudes (Meehl et al. 2005). Water supplies stored in glaciers and snow cover are projected to decline in the course of the century, thus reducing water availability during warm and dry periods (through a seasonal shift in streamflow, an increase in the ratio of winter to annual flows, and reductions in low flows) in regions supplied by melting water from major mountain ranges, where more than one-sixth of the world’s population currently lives (Stern 2007). Globally, the negative impacts of future climate change on freshwater systems are expected to outweigh the benefits (Vörösmarty et al. 2000; Alcamo et al. 2003; 2007; Arnell, 2004). By the 2050s, the area of land subject to increasing water stress due to climate change is projected to be more than double that with decreasing water stress. Areas in which runoff is projected to decline face a clear reduction in the value of the services provided by water resources. Increased annual runoff in some areas is projected to lead to increased total water supply. However, in many regions, this benefit is likely to be counterbalanced by the negative effects of increased precipitation variability and seasonal runoff shifts in water supply, water quality and flood risks (Bates et al. 2008). Climate change affects the function and operation of existing water infrastructure – including hydropower, structural flood defences (Maris et al. 2015), and drainage and irrigation systems – as well as water management practices (Schudder 2005). Adverse effects of climate change on freshwater systems aggravate the impacts of other stresses, such as population growth, changing economic activity, land-use change and urbanisation. Globally, water demand will grow in the coming decades, primarily due to population growth

29 and increasing affluence; regionally, large changes in irrigation water demand as a result of climate change are expected (Vörösmarty et al. 2010). Current water management practices may not be robust enough to cope with the impacts of climate change on water supply reliability, flood risk, health, agriculture, energy and aquatic ecosystems. In many locations, water management cannot satisfactorily cope even with the current climate variability, so that large flood and drought damages occur. As a first step, improved incorporation of information about current climate variability into water- related management would assist adaptation to longer-term climate change impacts. Climatic and non-climatic factors, such as growth of population and damage potential, would exacerbate problems in the future (Bates et al. 2008). Climate change challenges the traditional assumption that past hydrological experience provides a good guide to future conditions. The consequences of climate change may alter the reliability of current water management systems and water-related infrastructure. While quantitative projections of changes in precipitation, river flows and water levels at the river-basin scale are uncertain, it is very likely that hydrological characteristics will change in the future. Adaptation procedures and risk management practices that incorporate projected hydrological changes with related uncertainties are being developed in some countries and regions (Barrios et al. 2009; Gujja et al. 2009; Kashaigili et al. 2009; Yu et al. 2009). Adaptation options designed to ensure water supply during average and drought conditions require integrated demand-side as well as supply-side strategies. The former improve water-use efficiency, e.g., by recycling water. An expanded use of economic incentives, including metering and pricing, to encourage water conservation and development of water markets and implementation of virtual water trade, holds considerable promise for water savings and the reallocation of water to highly valued uses. Supply-side strategies generally involve increases in storage capacity, abstraction from water courses, and water transfers. Integrated water resources management provides an important framework to achieve adaptation measures across socio-economic, environmental and administrative systems. To be effective, integrated approaches must occur at the appropriate scales (Moench et al. 2003). Mitigation measures can reduce the magnitude of impacts of global warming on water resources, in turn reducing adaptation needs. However, they can have considerable negative side effects, such as increased water requirements for afforestation/reforestation activities (Calder 1990) or bio-energy crops (IPCC 2000), if projects are not sustainably located, designed and managed. Water resources management clearly impacts on many other policy areas, e.g. energy, health, food security and nature conservation. Thus, the appraisal of adaptation and

30 mitigation options needs to be conducted across multiple water-dependent sectors. Low- income countries and regions are likely to remain vulnerable over the medium term, with fewer options than high income countries for adapting to climate change. Therefore, adaptation strategies should be designed in the context of development, environment and health policies (Parry et al. 2007). Several gaps in knowledge exist in terms of observations and research needs related to climate change and water (Dai and Trenberth 2002). Observational data and data access are prerequisites for adaptive management, yet many observational networks are shrinking. There is a need to improve understanding and modelling of climate changes related to the hydrological cycle at scales relevant to decision making (Frisch et al. 2002; GCOS 2003; 2004). Finally, information about the water related impacts of climate change is still inadequate – especially with respect to water quality, aquatic ecosystems and groundwater – including their socio-economic dimensions (IPCC 2007; Bates et al. 2008).

2.6 Water resources, climate change and economic impacts

Climate and water resources impact on several secondary and tertiary sectors of the economy such as insurance, industry, tourism and transportation. Water related effects of climate change in these sectors can be positive as well as negative, but extreme climate events and other abrupt changes tend to affect human systems more severely than gradual change, partly because they offer less time for adaptation (Bates et al. 2008). Industrial sectors are generally thought to be less vulnerable to the impacts of climate change than such sectors as agriculture. Among the major exceptions are industrial facilities located in climate-sensitive areas (such as floodplains) (Ruth et al. 2004). For the finance sector, climate-change related risks are increasingly considered for specific 'susceptible' sectors such as hydro-electric projects, irrigation and agriculture, and tourism (UNEP/GRID-Arendal 2002). Effects of climate change on tourism include changes in the availability of water, which could be positive or negative (Braun et al. 1999; Uyarra et al. 2005). Warmer climates open up the possibility of extending 'exotic' environments (such as palm trees forest in Crete Island in southern Mediterranean), which could be considered by some tourists as positive but could lead to a spatial extension and amplification of water- and vector-borne diseases. Droughts and the extension of arid environments (and the effects of extreme weather events) might discourage tourists, although it is not entirely clear what they consider to be unacceptable (IPCC 2007).

31

2.7 Water resources, climate change and agriculture

Water is a central issue of adaptation to climate change in agriculture. Agricultural production depends critically on how climatic variables such as precipitation, evapotranspiration and temperatures vary across regions and over time. The effects of climate change on agriculture occur through crop water requirements, availability and quality of water, and other factors, which are affected by both long-term gradual change and extreme events, and across a range of scales from local to regional to continental. Moreover, climate is not only changing but is becoming non-stationary, meaning that expectations can no longer be based only on past observations. Interactions between climate change, water and agriculture are numerous, complex and region-specific. Climate change can affect water resources through several dimensions simultaneously:

. Changes in the amount and time patterns of precipitation . Impact on water quality through changes in runoff . Impact on river flow . Impact on retention and thus loading of nutrients . Extreme events such as flooding phenomena and drought conditions

Water supply can be directly impacted by climate change through changes in rainfall patterns, and indirectly through changes in water compartments such as surface water, groundwater, snow and glaciers that can be used for the purpose of agricultural water withdrawals, including irrigation and livestock (OECD 2014). Regarding irrigation, most plants need water for healthy growth, and water shortages decrease plant yields and increase plant stress and susceptibility. Many countries suffer from water shortages, and complex irrigation schemes have been developed to supplement water supplies to dry agricultural areas by means of dams and artificial channels. Irrigation agriculture is traditional form of agriculture in the Mediterranean countries and it is by far the biggest agricultural use of freshwater, and it grows bigger over time as the area under irrigation continues to expand (Park 2001). Nevertheless, different crops and regions vary in their water requirements. Additionally, coupled with inefficient crop choice, some irrigation practices waste large amount of water. The continued expansion of irrigation agriculture is causing mounting concern about the sustainability of water supplied in many areas. Rainfall patterns, temperature, evapotranspiration, soil quality and vegetative cover all influence soil moisture levels (Pimentel et al. 1997).

32

Interactions between relevant weather variables that affect agricultural production are difficult to characterize. Scientific evidence underlying projected impacts on freshwater has significant limitations when it comes to informing practical, on-site adaptation decisions. These complex interactions multiply the uncertainties concerning the impact of climate change on agriculture. The frequency and severity of extreme events such as flooding phenomena and drought conditions may increase as a result of climate change and have substantial negative impacts on agricultural production. Much of the work undertaken on the potential impact of climate change has focused on projected changes in average temperature and rainfall, and links between these changes and measurable outcomes with clear economic implications. Despite the low level of certainty concerning the scientific evidence regarding shifts in extreme events, non-linear (convex) damage functions mean that changes in extremes are expected to be most costly. Because agricultural water management involves public good, externalities and risk management issues, private adaptation to climate change is not equal to collective adaptation. A consistent strategy for agricultural water management needs to consider the following five levels of actions and their linkages:

. On-farm: Adaptation of water management practices and cropping and livestock systems . Watershed: Adaptation of water supply and demand policies in agriculture and with the other water users (urban and industrial) and uses (ecosystems) . Risk management: Adaptation of risk management systems against flooding phenomena and drought conditions . Agricultural policies and markets: Adaptation of existing agricultural policies and markets to the changing climate . Interactions: between mitigation and adaptation of agricultural water management

At the watershed level, well-designed, flexible and robust water sharing rules and economic instruments such as water pricing and water trading can foster adaptation of water systems. As climate is becoming non-stationary and climate risks are projected to increase, systems that allocate water across farms and across other uses should be flexible and robust enough to allow for efficient use of water, taking into account redistributive consequences and priority uses. Climate change adaptation and mitigation practices may have positive or negative implications on agricultural water management and on water quality. The potential synergies and trade-offs between mitigation and agricultural management practises are, however, site-

33 specific and for many cases there are substantial knowledge gaps. Although this is a complex matter, it is important to recognise these linkages in the design of mitigation policies to reduce the risk of conflict between mitigation and water policy objectives and to maximize potential synergies (OECD 2014).

2.8 Water resources in Greece

Water resources refer to the water 'produced' inside a country in conjunction with the external water contribution from neighbouring countries (e.g. trans-boundary rivers). According to data from the AQUASTAT (FAO 2016), the mean precipitation in Greece in 652mm/year while in Europe is 664mm/year and in the Mediterranean European countries is 768mm/year. In Greece, the western part accepts the majority of rainfalls with more than 1500mm/year, while eastern Greece, along with the islands of the Aegean and Crete, have considerably small rainfalls. The shortage of water (drought) in a region is not only related to the availability of the water resources, but also to the water utilization. Table 2.03 and figure 2.02 depict the supply-demand water situation for the 14 water districts in Greece (109 m3/year).

Table 2.03. Supply-demand water situation for the 14 water districts in Greece

No. Water Regions Supply Demand Status 1 West Peloponnisos 73 55 Surplus 2 North Peloponnisos 122 104 Surplus 3 East Peloponnisos 56 67 Deficit 4 West Central Greece 415 82 Surplus 5 Epirus 193 33 Surplus 6 Attica 56 54 Marginally Surplus 7 East Central Greece 128 187 Deficit 8 Thessaly 210 335 Deficit 9 West Macedonia 159 136 Surplus 10 Central Macedonia 137 130 Marginally Surplus 11 East Macedonia 354 132 Surplus 12 Thrace 424 253 Surplus 13 Crete 130 133 Marginally Deficit 14 Aegean Islands 7 25 Deficit Total 2.464 1.726 Surplus

34

450 415 424 400 Supply 354 350 Demand 335 300 253 250 210 193 200 159 137 150 122 187 132 133 82 100 73 67 128 136 130 130 104 56 50 33 25 55 56 54 7 0

Figure 2.02. Supply-demand water situation for the 14 water districts in Greece

Although it is evident that water supply-demand situation in Greece is presented with surplus, nevertheless, the situation is expected to become worse under a systematic climate change due to the greenhouse effect (Mimikou 2005). According to figure 2.02, the Ardas River basin belongs to the Thrace Region and it is the only one from the examined areas where the status regarding the water resources shows a surplus. On the other hand the Sperchios (East Central Greece) and the Geropotamos (Crete) River basins belong in water Regions that face deficit and marginally deficit conditions, respectively. In the era of climate change, it is useful to study the future climate change and their impact to the water resources as well as to assess the future responses of the aspects of the hydrological cycle in relation with the agriculture production, towards the sustainable water management and the sustainable development of the environment and the agriculture productivity.

35

36

Chapter 3

Data and Methodology

3.1 Study areas

The study areas are located in Greece. The Ardas River basin (Study area I) is located in the north-eastern part of Greece (a significant part of the watershed belongs to Bulgaria - mostly the mountainous area of the basin where the springs of Ardas River are located). The Sperchios River basin (Study area II) is located in Central Greece, while the Geropotamos River basin (Study area III) is located in Crete Island in southern Greece. Figure 3.01 depicts the general location of the study areas, while table 3.01 presents the local population that is employed in the primary section, or are related with agricultural activities, as well as the main characteristics of the study areas.

Table 3.01. Characteristics of the study areas

Study areas Ardas Sperchios Geropotamos Location North-eastern Greece Central Greece South Greece Region Evros Ftiotida Irakleion Region population (inhabitants) 149.354 169.542 304.270 People occupied in the primary sector 37.560 52.426 116.251 Percentage (%) 25.1 30.9 38.2 Climate Classification (Köppen-Geiger) Cfa, Cfb, Dfb Csa, Csb Csa Area (km2) 5681.3 1727.7 651.6 Perimeter (km) 513 259.2 113.3 Mean Slope (%) 4.8 13.1 8.7 Elevation range (m) 1 - 2081 0 - 2277 0 - 2380 Mean annual precipitation (mm) 839.8 792.9 759.8 Mean annual temperature (oC) 10.5 16.6 17.2 Length of Hydrographic Network (km) 8907.8 4519.9 1038.8

37

Figure 3.01. Location and characteristics of the study areas 38

Study area I - Ardas River basin Table 3.02. Meteorological stations and characteristics - Ardas River basin Station Elevation (m) Lat. (o) Long. (o) Years Variables* Dikaia 90.7 41.70 26.30 PREC

Kuprinos 51.2 41.58 26.23 PREC

Sitoxori 112.4 41.46 26.35 PREC

Mikro Dereio 129.2 41.31 26.10 PREC Megalo Dereio 401.4 41.23 26.01 PREC Protoklisi 55.1 41.30 26.25 PREC 1985-2000 Didimoteixo 48.4 41.35 26.50 PREC

Metaxades 129.0 41.41 26.23 PREC; T; Rs; RH; WS Orestiada 39.2 41.50 26.51 PREC; T; R ; RH; WS s Edirne 50.0 41.66 26.56 PREC; T; RH;

Kurdjali 273.3 41.65 25.36 PREC; T; RH; Figure 3.02. Meteorological stations - Ardas River basin Ivaylovgrad 120.5 41.58 26.10 Q

32.6% 243 - Land principally covered by agriculture

21.5% 311 - Broad-leaved forest

9.3% 312 - Coniferous forest

17.8% 313 - Mixed forest 3.3% 321 - Natural Grasslands

0.2% 323 - Sclerophylous vegetation

15.3% 324 - Transitional woodland shrub

o 2 *PREC=Precipitation (mm); T=Tmean, Tmax, Tmin ( C); Rs= solar radiation (MJ/m ); Figure 3.03. Land Use - Ardas River basin 3 RH=Relative Humidity (%); WS=Wind Speed (m/s); Q=Discharge (m/s )

39

Study area II - Sperchios River basin Table 3.03. Meteorological stations and characteristics - Sperchios River basin Station Elevation (m) Lat. (o) Long. (o) Years Variables* Neoxori 800 38.96 21.86 PREC

Pitsiota 800 39.01 21.9 PREC

Zileuto 120 38.93 22.26 PREC

Trilofos 580 39 22.21 PREC Timfristos 850 38.91 21.91 PREC Ypati 286 38.86 22.23 PREC Duo Vouna 460 38.79 22.38 1981-2000 PREC

Lamia 144 38.91 22.43 PREC; T; Rs; RH; WS Karpenisi 819 38.9 21.78 T; R ; RH; WS s Domokos 522 39.13 22.3 T; R RH; WS s Kastri 100 38.94 22.2 Q Figure 3.04. Meteorological stations - Sperchios River basin Kompotades 20 38.86 22.36 Q Komma 12 38.84 22.43 Q

24.7% 243 - Land principally covered by agriculture

12.1% 311 - Broad-leaved forest

12.0% 312 - Coniferous forest

9.5% 313 - Mixed forest 1.6% 321 - Natural Grasslands

23.6% 323 - Sclerophylous vegetation

16.5% 324 - Transitional woodland shrub

*PREC=Precipitation (mm); T=T , T , T (oC); R = solar radiation (MJ/m2); Figure 3.05. Land Use - Sperchios River basin mean max min s RH=Relative Humidity (%); WS=Wind Speed (m/s); Q=Discharge (m/s3)

40

Study area III - Geropotamos River basin Table 3.04. Meteorological stations and characteristics - Geropotamos River basin Station Elevation (m) Lat. (o) Long. (o) Years Variables*

Souda 106.4 35.54 24.10 PREC; T; Rs; RH; WS

Irakleio 68.3 35.32 25.17 PREC; T; R ; RH; WS s Ierapetra 24.2 35.01 25.72 PREC; T; R ; RH; WS s Siteia 25 35.19 26.09 PREC; T; Rs; RH; WS Rethimno 118 35.34 24.50 PREC; T; Rs; RH; WS Tympaki 33.7 34.99 24.74 PREC; T; Rs; RH; WS

Palaioxora 25 35.23 23.68 1981-2000 PREC; T; Rs; RH

Anogeia 823.7 35.28 24.95 PREC; T; Rs; RH Fourni 500 35.25 25.66 PREC; T; R ; RH s Kastelli 350 35.12 25.20 PREC; T; R ; RH; WS s Zaros 322 35.13 24.90 PREC; T; Rs; RH Figure 3.06. Meteorological stations - Geropotamos River basin Faistos 95.7 35.04 24.79 Q

Souda 106.4 35.54 24.10 PREC; T; Rs; RH; WS

70.4% 243 - Land principally covered by agriculture

0.1% 311 - Broad-leaved forest

1.2% 312 - Coniferous forest

0% 313 - Mixed forest 13.1% 321 - Natural Grasslands

10.1% 323 - Sclerophylous vegetation

5.1% 324 - Transitional woodland shrub

*PREC=Precipitation (mm); T=T , T , T (oC); R = solar radiation (MJ/m2); Figure 3.07. Land Use - Geropotamos River basin mean max min s RH=Relative Humidity (%); WS=Wind Speed (m/s); Q=Discharge (m/s3)

41

3.2 Climate

The study areas present different climate characteristics. The Ardas River basin is characterised by the existence of various climate types, but mainly faces humid continental conditions. In the summer period, the Ardas River basin faces warm and in some parts hot conditions, while the winter period is mild to cool. Precipitation falls throughout the year. During the winter season, snowfall occurs in many areas and snow cover is often deep. Regarding the climate conditions in the Sperchios River basin, mostly the biggest part of the basin faces Mediterranean conditions with hot-summer Mediterranean climate. Some minor variations appear on the Rivers' springs on the mountainous area where milder conditions appear with warm-summer Mediterranean climate. Generally, the summer season is characterised by insignificant amount of precipitation and very warm to hot conditions, while the winters are mild with snowfalls in the highlands that The Geropotamos River basin is characterized by purely Mediterranean conditions with hot-summer Mediterranean climate. The summer season presents a general dryness, while the winters are mild to cool, and wet. The Geropotamos River basin receives almost all its amount of precipitation during winter, autumn, and spring seasons, and it can go up to 4-6 months during the summer period without any significant precipitation. The drier atmospheric conditions are a result of strong, desiccating winds on the leeward side of the mountainous areas (Peel et al. 2007). Figures 3.08, 3.09, and 3.10 depict the Köppen-Geiger classification in the examined study areas.

42

Figure 3.08. Köppen-Geiger classification - Ardas River basin

Figure 3.09. Köppen-Geiger classification - Sperchios River basin

Figure 3.10. Köppen-Geiger classification - Geropotamos River basin

43

3.3 Climate Data

Daily values of meteorological data are obtained from the stations within or adjacent to the study areas as can be seen in figures 3.02, 3.04 and 3.06, derived from the Hellenic National Meteorological Service (HNMS) and OGIMET1 are used in the study. Future data are obtained by the output simulations of an ensemble of six Regional Climate Models (RCMs), which were carried out within the European project ENSEMBLES. The characteristics of the RCMs and the driving Global Circulation Models (GCMs) can be found in table 3.05.

Table 3.05. Driving GCMs, Institutes, RCMs, and relevant references of the used ENSEMBLES simulations

Driving GCM Institute RCM References Norwegian Meteorological METNO_HIRHA Christensen et al. BCCR-BCM2 Institute M (1996) Centre National de Gibelin and Deque CNRM-CM3 Recherches CNRM-RM4.5 (2003) Meteorologiques Danish Meteorological Jungclaus et al. (2006) ECHAM/MPI-OM1 HIRHAM5 Institute Roeckner et al. (2006) METO-HC- Met Office Hadley Centre, HADCM3C Johns (2009a) HadCM3C UK METO-HC- Met Office Hadley Centre, HADGEM2 Johns (2009b) HadGEM2 UK HadCMeQ0 (normal Swiss Federal Institute of ETHZ-CLM Jaeger et al. ( 2008) sensitivity) Technology

The climate simulations concern the future periods 2021-2050 and 2071-2100 against the reference periods that were pre-selected for each study area, under the A1B and B1 IPCC climate forcing scenarios. According to the Intergovernmental Panel on Climate Change (IPCC 2007), climate forcing refers to anything which forces a change on the climate system. In the specific case, it refers to the emission scenarios or representative concentration pathways, that are providing projections regarding the atmospheric concentrations of the greenhouse gasses. These scenarios are used as the driving inputs to the global climate models, as the

1 www.ogimet.com 44 greenhouse gas concentrations influence the balance between the incoming and the outgoing radiation. Thus, each set of climate projections is valid under a specific assumption of future anthropogenic greenhouse gas concentrations. There are three main sets of climate change forcing scenarios: SRES scenarios, non- SRES scenarios, and the latest RCP scenarios. These are forty SRES scenarios grouped into four families (A1, B1, A2, B2), based on narratives of demographic, social, economic, technological, and environmental development. There are five widely used illustrative scenarios: A1B, A1F1, A1T, A2, and B1. Table 3.06 presents a short description of the scenarios that were used in the current study (namely A1B and B1).

Table 3.06. IPCC emission scenarios (IPCC 2001)

A future world of very rapid economic growth, low population growth and rapid introduction of new and more efficient technology. Major underlying themes are economic and cultural convergence and capacity building, with a substantial A1B reduction in regional differences in per capita income. In this world, people pursue personal wealth rather than environmental quality. Energy technologies balanced across energy sources. A convergent world with the same global population as in the A1 storylines but with rapid changes in economic structures toward a service and information economy, B1 with reductions in materials intensity, and the introduction of clean and resource- efficient technologies.

3.4 Data homogeneity, correlation and future data extraction

Regarding the reference period data for the study areas, since there are no available common meteorological data series, a homogeneity test is performed, followed by a statistical correlation test, to correlate the existing values and determine the reference periods that will be adopted. The technique of data homogeneity was conducted using the double-mass curve, proposed by Dingman (1994; 2002). Since the existence of limited data in Greece is an issue that the researcher is called to overcome (Paparrizos et al. 2014), homogeneity check is one of the first steps and essential tool in the analysis of a long record. The double-mass curve is used to check the consistency of many kinds of hydrological and meteorological data by comparing the data for a single station with that of a pattern composed of the data from several other stations in the area. It constitutes a commonly used data analysis approach for investigating the behaviour of various hydrometeorological records and it is often used to determine whether there is a

45 need for corrections to the data - to account for changes in data collection procedures or other local conditions. Such changes may often result from a variety of things including changes in instrumentation, changes in observation procedures, or changes in gauge location or surrounding conditions (Dingman 1994; 2002; Paparrizos and Chatziminiadis 2010; Paparrizos 2012). Additionally, the correlation of the existing data series is performed using the t-test (Snedecor and Cochran 1989; Haan 2002). Many tests have been proposed in the literature for the statistical significance of correlation between different data series. In the current study, due to the fact that the sample of data (daily values transformed into monthly and finally yearly values) is relatively small (<25), the t-test is preferred. The significance of a sample correlation, r, depends on the sample size and also on the size of r. The assumptions according to t-test are:

. The samples, x and y, are drawn from populations that follow a bivariate normal distribution

. The samples constitute random samples from the population

. The population correlation coefficient is zero: ρ = 0

If these assumptions are satisfied (which they are in the certain study), then according to equation 3.01:

(3.01)

the statistic equation follows a t-distribution with N - 2 degrees of freedom, where:

N is the sample size r is the correlation coefficient between these two samples

The null and alternative hypotheses for a test (two-sided) are:

H0: ρ = 0 H1: ρ ≠ 0

46

Combining the t-test and the double-mass curve homogeneity test, the periods that were determined for the study areas were 1981-2000 for the Sperchios and the Geropotamos River basins, and 1985-2000 for the Ardas River basin. Regarding the future simulated data that they are occurred by the output simulations of the European project ENSEMBLES, their original form was network Common Data Form (netCDF) files. NetCDF files is a set of software libraries and self-describing, machine- independent data formats that support the creation, access, and sharing of array-oriented scientific data (Open Geospatial Consortium 2016). The commercial software package MATLAB2014a (MATLAB 2014a, The Mathworks Inc., Natick, MA, 2000) was used to access the data through the 'MATLAB netcdf' package which provides interfaces to dozens of functions in the netCDF library. Additionally, a script was developed to extract the specific sets of data according to the location of the study areas. An indicative initial ‘read’ of an ENSEMBLES data series can be found in Appendix A1, while the developed script in order to extract the necessary data can be found in Appendix A2.

3.5 Downscaling and spatial interpolation techniques

One of the biggest issues faced by atmospheric scientists nowadays is the accurate representation and prediction of various climatological parameters. In this direction, one of the most important observational tools that climatologists have in their service in order to collect data is the radar. Its applicability has been tested since the 1940's, mostly to detect and forecast extreme weather conditions, but also to contribute to the accurate knowledge of the climatic variations prevailing in and dominating in a certain area (Maris et al. 2013). In regions, however where radar grid observations are not available and the meteorological network does not have the density and thus is insufficient and unable to cover the whole area of study, or when performing a future assessment of the spatial distribution of the climatic variables and hence there are no available radar grid observations, downscaling through spatial interpolation is mandatory. A variety of interpolation methods have been developed for climatological data mapping. Most of them are based mainly on the similarity and topological relations of nearby sample points and on the value of the variable to be measured (Beek et al. 1992; Chang et al. 2005; Gemmer et al. 2004; Maris et al. 2013). Geostatistical interpolation has become an important tool in applied climatology because it is based on the spatial variability of the variables of interest and makes it possible to quantify the estimation uncertainty (Gambolati and Volpi 1979; Chua and Bras 1982; Myers 1982; Bacchi and Kottegoda 1995; Martinez- Cob 1996; Holawe and Dutter 1999; Paparrizos et al. 2015). In this latter case, the General Circulation Models, which according to Tolika et al. (2008) although they remain, nowadays,

47 the most appropriate tool for the development of future climate scenarios, nevertheless, they operate in macro-scale. The need for regional projections of the changes in extreme meteorological parameters as well as the mismatch between spatial scale and the climate impact models - which operate on the meteorological mesoscale (Schubert and Henderson-Sellers 1997), lead to a wide development of several downscaling techniques. These downscaling processes are divided in two regional subcategories:

. the dynamical approach, employing regional climate models (RCMs)

. the statistical approach, where empirical-statistical relationships are defined between the independent variables (predictors) and the dependent variables (predictants)

Although RCMs are considered to be more promising at the evolution of downscaling, the statistical models present some advantages, which make them useful to the researcher (Goodess and Palutikof 1998; Timbal et al. 2003). Thus, in order to begin with the downscaling and spatial interpolation, a variable needs to be selected first, that will be downscaled and spatially interpolated in a certain study area. The initial climatological data will be obtained from the climatological stations (point stations) within or adjacent to the examined area. While mapping a climatological variable in a certain area, several factors that affect the certain variable need to be taken into account. In order to depict the relationships of this examined variable with the various factors that affect it, a multi-linear regression technique needs to be performed including information from the climatological stations of the study area, using as dependent factor the variable that needs to be spatially interpolated, and as independent all the factors that affect this specific variable. Following that procedure, an equation (Eq. 3.02) will be created for this certain variable and study area:

(3.02)

Where: x represents the dependent variable at a certain point (climatological variable value) bo is constant from the multi-linear regression equation b1...bn represent the coefficients obtained for each independent factor from the multi-linear regression equation

48 a...m represent the factors that were selected to be used in the multi-linear regression procedure and affect each time the certain climatological variable (e.g. elevation, longitude, latitude, slope, aspect, other meteorological variables, distance from a water body, etc.)

During the multi-linear regression and in order to observe the statistical significance of the examined factors, special attention needs to be given in the output p-value. P-value indicates the probability of obtaining a test statistic result at least as extreme as the one that was actually observed, assuming that the null hypothesis is true (Pearson 1900). P-value also determines which factors will be used in the multi-linear regression in order to eliminate deficiencies. The preconditions that the output p-value should meet are:

. The output p-value should belong each time within the significance level (p≤ 0.05), in order for a strong presumption to exist against the null hypothesis

. In cases where the significance level of the examined factor is less than 95% during the multi-linear regression analysis, this factor should be eliminated from the procedure, and the multi-linear regression needs to be re-performed

Table 3.07 presents the categorization of the output p-values that has been made in accordance with the significance level of 95% (p≤ 0.05). A noteworthy fact is that the selection of the factors that will be included each time in the regression procedure is not determined as each climatological variable is influenced by various factors, thus the selection of the factors that will be used every time lies within the responsibilities of the researcher and the availability of the data. This constitutes the statistical approach.

Table 3.07. Categorization of the output p-values at 95% significance level (p≤ 0.05)

p ≤ 0.01 very strong presumption against null hypothesis 0.01 ≤ p ≤ 0.05 strong presumption against null hypothesis 0.05 ≤ p ≤ 0.1 low presumption against null hypothesis p > 0.1 no presumption against null hypothesis

The methodology that will be applied to the interpolation from irregularly distributed surface station data at coordinates xi, yi, zi (where xi = longitude, yi = latitude and zi = altitude from mean sea level of the i station) to surface gridded points Xj, Yj, Zj (where Xj = longitude,

Yj = latitude and Zj = altitude from mean sea level of the j grid) is based on the Ordinary

49

(spherical variogram) Kriging with the procedure of interpolation for geographic information systems (Oliver and Webster 1990) through ArcGIS 10.2.1 program. Subsequently, through ArcGIS 10.2.1, in order to perform the Ordinary Kriging analysis, automatic points (x) within the basin will be created (through Fishnet option from the Data Management Toolbox menu) in a 1x1km resolution grid and through the multi-linear regression equation a value will be given (according to the dependent variable) to every point by feeding them each time with different values from the independent factors. This constitutes the dynamical approach of the downscaling technique. Figure 3.11 gives a schematic representation and it is rather indicative of the procedure that was described above.

Figure 3.11. Dynamical and statistical downscaling procedure 50

3.6 Trend analysis of climatological data

Trend analysis is the practice of collecting information and attempting to spot a pattern, or trend, in the information. In some fields of study, the term 'trend analysis' has more formally defined meanings (Immerwahr 2004). Although trend analysis is often used to predict future events, it could be used to estimate uncertain events in the past. This detection, estimation, and prediction of trends and associated statistical and physical significance are important aspects of climate research (Salmi et al. 2002). In the current study, trend analysis is performed using the Mann-Kendall test for the reference years that were defined after the data homogeneity and correlation, and for up to 2100 according to the future periods that were defined. The Mann-Kendall test (Kendall 1938; Mann 1945) is a trend analysis test, suitable for cases where the trend may be assumed to be monotonic and thus no seasonal or other cycle is present in the data. The

Mann-Kendall test is applicable in cases where the data values xi of a times series can be assumed to obey the model that has the following form (Eq. 3.03):

(3.03)

Where: xi represents the data values of a time series that assume to obey the model f(t) is a continuous monotonic increasing or decreasing function of time

εi is the residual that can be assummed to be from the same distribution with zero mean

It is therefore assummed that the variance of the distribution is constant in time. For time series with less than 10 data points the S test is used, and for time series with 10 or more data points, the normal approximation is used. Since in the current study the number of data values that are being tested regarding their trend sensitivity (mean annual values) are each time 10 or more (n ≥ 10), the normal approximation will be used, and the variance of S will be first computed by the following equation (Eq. 3.04) which takes into account that ties may be present:

(3.04)

Where: n is the number of annual values in the studied data series q is the number of tied groups th tp is the number of data values in the p group 51

The values of S and VAR(S) are used to compute the test statistic Z as follows (Eq. 3.05):

(Eq. 3.05)

The presence of a statistically significant trend is evaluated using the Z value. A positive (negative) Z value indicates an upward (downward) trend. The statistic Z value has a normal distribution.

3.7 Potential Evapotranspiration formulae

The general equipment used to measure ET includes the lysimeter, Bower-ratio energy balance system, eddy covariance technique (ECT), scintillometers, etc. Nonetheless, in cases where the necessary equipment for the measurement of ET is not available, ET can be estimated by theoretical or empirical equations that require simple or analytical data (Kotsopoulos and Babajimopoulos 1997). Knowledge of PET rates is essential for a variety of applications including assessment of the hydrological water balance modeling, water resources planning and management, irrigation scheduling and planning, agricultural water use, geo-botanical studies, and estimation of sensitive-to-climate change, aridity indices (Valiantzas 2006). In the current study 12 Potential Evapotranspiration (PET) formulae are used that include the approaches of Hargreaves (Hargreaves and Samani 1985; Allen et al. 1998; Xu and Singh 2002; Oudin et al. 2005), Priestley-Taylor (Priestley and Taylor 1972; Winter et al. 1995; Yao 2009), Hamon version 1 (Lu et al. 2005), Hamon version 2 (Oudin et al. 2005), Hamon version 3 (Rosenberry et al. 2004), McGuiness (Oudin et al. 2005), Romanenko (Oudin et al. 2005), Caprio (Caprio 1974), Jensen-Haise (Jensen and Haise 1963; Rosenberry et al. 2004), Turc (Jacobs and Satti 2001; Lu et al. 2005), Ivanov (Wendling et al. 1984) and the ASCE Penman-Monteith equation (ASCE Task Committee 2005). The equations are presented in table 3.08, while the input data requirements (and thus the variables that affect every equation and are analyzed regarding their sensitivity) are presented in table 3.09. The equations include a wide variety of input parameters thus their applicability depends each time on the data availability.

52

Table 3.08. Potential Evapotranspiration formulae Common name Mathematical equation

Hargreaves

Priestley-Taylor

Hamon 1

Hamon 2

Hamon 3

McGuiness

Romanenko

Caprio

Jensen-Haise

Turc

Ivanov

ASCE

o o Note: RH = Relative air humidity (%); Tmean = mean air temperature ( C); Tmax = maximum air temperature ( C); Tmin = minimum air o o -2 temperature ( C); Tdew = dew point temperature ( C); Rs = total global solar radiation (MJ m ); Ra = total solar radiation reaching the -2 -2 -1 atmosphere's surface (MJ m ); Rn = total net solar radiation (MJ m ); λ = latent heat of vaporization (MJ ∙ kg ); α = proportionally coefficient (unitless - set as 1.2 in the current study); SVD = saturated vapour density at mean air temperature (g ∙ m-3); Δ = the slope of saturation vapour pressure curve at air temperature T (kPa oC -1); G = the soil heat flux density at the crop surface (MJ m-2); for daily -1 o -1 periods the value is 0); U2 = the mean wind speed at 2 m height (m ∙ s ); γ = the psychrometric constant (kPa C ); es = the mean saturation vapour pressure at 1.5-2.5 m height (kPa); ea = the mean actual vapour pressure at 1.5-2.5 m height (kPa); and Cn = a 3 -1 numerator constant that changes with reference type. Cn takes the value 900 and 1600 (K mm s Mg ) for short and tall crops, -1 -1 respectively; Cd = a denominator constant that changes with reference type and takes the values of 0.34 (s m ) and 0.38 (s m ) for short and tall crops, respectively; αPT = Priestley-Taylor coefficient (unitless - set as 1.26 in the current study); DL = Day Length (hours). 53

Table 3.09. Input data requirements of used PET formulae

Tmean Tmin Tmax Tdew RH U2 DL Rs Ra Rn

oC oC oC oC % m/s h. MJ/m2 MJ/m2 MJ/m2

PETHar    

PETPT 

PETHam1    

PETHam2  

PETHam3    

PETMcG  

PETRom    

PETCap  

PETJen  

PETTur  

PETIva  

PETASCE       Note: Detailed explanation of the abbreviations of the variables in the current table is given below Table 3.08.

3.8 Sensitivity analysis

Sensitivity analysis investigates the effect of change of one factor on another (McGuen 1973). In many cases it is usually the first step towards model calibration because it answers several questions such as where data collection efforts should focus, what degree of care should be taken for parameter estimation, and which is the relative importance of various parameters (Cho and Lee 2001). A sensitivity analysis also identifies the most sensitive parameters, which ultimately dictates the set of parameters to be used in the subsequent calibration process. There are different methods available for carrying out and expressing their results (Lenhart et al. 2002; Van Griensven et al. 2002; 2006). Some methods use a percentage change in input and respond to corresponding changes in output variables, while some other methods use an increase or decrease in certain proportion and record the observed changes (Kannan et al 2007). In hydrometeorological studies and ecological applications, a number of sensitivity coefficients have been defined depending on the purpose of the analyses (Coleman and DeCoursey 1976; Beven 1979; Beres and Hawkins 2001; Anderton et al. 2002). Literature reviews of previous studies revealed that there is no standard or common procedure for computing sensitivity coefficients for climate variables (Irmak et al. 2006; Estevez et al. 54

2009). In the current study, sensitivity analysis is performed using a new coefficient proposed by Ampas (2010), and is presented in equation 3.06:

(3.06)

Where:

Ksp is the new coefficient p is the examined independent variable or parameter M is the modelled value

σp is the standard deviation of the meteorological variable's data series

The current equation presents several advantages compared to other approached, which are the following (Ampas and Baltas 2012):

. Standard deviation cannot be zero

. The coefficient is not influenced by the units

. The minimum value depends on the magnitude of the time series

. The range width depends on both minimum and maximum values

. Some meteorological parameters like Relative Humidity and Wind Speed are limited,

with RH from 0 - 100%, while U2 is positive (> 0 m s-1)

. This sensitivity coefficient shows the alteration caused to the model by the usual change of the parameter

3.9 Aridity Index

Aridity is a term that most people conceptually understand, and it evokes images of dry, desert lands with sparse natural surface-water bodies and rainfall, and commonly only scant vegetation, which is adapted to a paucity of water (Maliva and Missimer 2012). Aridity, as defined by the shortage of moisture, is essentially a climatic phenomenon that is based on average climatic conditions over a region (Agnew and Anderson 1992). A fundamental distinction exists between aridity, which is a long-term climatic phenomenon and droughts, which are a temporary phenomenon (water deficit). 55

In the current study, Aridity Index (AI) which is delivered from the United Nations Environment Program (UNEP 1992) and constitutes a numerical indicator of the degree of dryness of a climate at a given location is selected to estimate the aridity conditions prevailing in the study areas. Aridity Index (AI) uses precipitation and potential evapotranspiration data, and it is given from the following equation (Eq. 3.07):

(3.07)

Where: P is the average annual precipitation (mm) PET is the average annual potential evapotranspiration (mm)

The boundaries that define various degrees of aridity are involved as shown in table 3.10 (FAO 1993).

Table 3.10. Classification of Aridity Index categories

Classification Aridity Index (AI) Hyper-arid AI ≤ 0.05 Arid 0.05 ≤ AI < 0.20 Semi-arid 0.20 ≤ AI < 0.50 Dry sub-humid 0.50 ≤ AI < 0.65 Sub-humid 0.65 ≤ AI < 0.80 Humid 0.80 ≤ AI < 1.5 Very humid 1.5 ≤ AI

3.10 Standardized Precipitation Index

Determination of drought concept has always been a brake on monitoring and analysis of the phenomenon. Additionally, more that 150 definitions have been attributed to drought in literature (Nastos et al. 2013a), and although there is no specific definition for drought, it is clear that it relates to periods of abnormally low water availability. These levels occur from the reduction of the frequency and the periods of rainfall (Bates et al. 2008). Decreased land precipitation and increased air temperature, which enhance evapotranspiration and reduce soil moisture, are important factors that have contributed to more regions experiencing droughts (Dai et al. 2004).

56

There are very few direct estimations of drought related variables, such as soil moisture, so drought proxies and hydrological drought proxies are often used to assess drought conditions. One of the major drought indices is the Standardized Precipitation Index (SPI), which is based on the probability of precipitation for any time scale. The SPI developed by McKee et al. (1993; 1995) is used in order to estimate the drought conditions prevailing in the study areas. SPI is an index that uses only precipitation data, it is based on the probability of precipitation for a number of consecutive months, and its main objective is to represent the deficit of precipitation over an area on multiple time scales relative to its climatology. Although SPI is not a drought prediction tool, the SPI methodology has been used to identify dryness and wetness conditions, and to evaluate their impact on water resources management. SPI can be computed in different running time intervals, i.e. 1-, 3-, 6-, 12-, and 24-months, but the index is flexible with respect to the chosen period (Vasileiades et al. 2009). This powerful feature can provide an overwhelming amount of information, unless researchers have a clear idea of the desired intervals (Karavitis et al. 2011). Mathematically, the SPI is based on the cumulative probability of a given rainfall event occurring at a station. The historic rainfall data of the station is fitted to a gamma distribution, as the gamma distribution has been found to fit the precipitation distribution quite well. This is done through a process of maximum likelihood estimation of the gamma distribution parameters, a and β. In simple terms, this process allows rainfall distribution at the station to be effectively represented by a mathematical cumulative probability function. Therefore, based on the historic rainfall data, the analyst can tell what is the probability of the rainfall being less than or equal to a certain amount. More information on the mathematical background of the SPI can be found in Edwards and McKee (1997). The classification scale for the SPI values is depicted in table 3.11.

Table 3.11. Classification scale for the SPI values

SPI value Category Probability (%) 2.00 or more Extremely wet 2.3 1.50 to 1.99 Severely wet 4.4 1.00 to 1.49 Moderately wet 9.2 -0.99 to 0.99 Near normal 68.2 -1.49 to -1.00 Moderately Dry 9.2 -1.99 to -1.50 Severely Dry 4.4 -2.00 or less Extremely Dry 2.3

57

3.11 Runoff assessment using ArcSWAT

Knowledge of the variation of future runoff in comparison with other fundamental aspects of the hydrological cycle under the prism of agricultural production in areas that present different climate characteristics is very essential. In that concept, ArcSWAT ArcGIS extension is used, based on the SWAT model, which is a powerful tool for the assessment of the water balance under the GIS spectrum. The ArcSWAT ArcGIS extension is a graphical user interface for the SWAT (Soil and Water Assessment Tool) model (Arnold et al. 1998). SWAT is a river basin, or watershed, scale model developed to predict the impact of land management practises on water, sediment, and agricultural chemical yields in large, complex watersheds with varying soils, land use, and management conditions over long periods of time. The model is physically based and computationally efficient, uses readily available inputs and enables users to study long-term impacts (Winchell et al. 2013). The calibration of ArcSWAT model includes the choice of parameters which each time play a key role in the natural process of converting rainfall into runoff. In the current study, the model is calibrated using the observed daily discharge data series that are collected from the stations that were used in the current study. In order to determine whether there is a successful calibration/verification of the model, the Coefficient of Efficiency (EF) or otherwise known as the Nash-Sutcliffe coefficient was used, whose equation (Eq. 3.08) has the following form:

n 2  (Oi  Pi ) EF  1.0  i1 n (3.08) 2  (Oi  O ) i1

Where:

Pi are the values of discharge predicted by the model

Oi are the values of observed (historical) discharge is the average of the historical discharge

EF is considered among the most important statistical indicators (Henriksen et al. 2003) and can take values in the period from - ∞ to 1.0, while the optimal value is 1.0. If it gets a value equal to 1.0 it means that the model describes with the same efficiency as the average of observed values, while negative values of the coefficient suggest that the applicability of the model is insufficient and needs to be re-calibrated.

58

Additionally, in an effort to further assess the future hydrological regime and the variations of the future water resources in comparison with the undergoing simulations of the current study, yearly data regarding the irrigation water withdrawal in Greece are collected from the AQUASTAT main data base, provided by the FAO (FAO 2015) for the years 1988- 2012. The data are further forecasted using the autoregressive integrated moving average (ARIMA) model method within Mathworks 2014a environment up to year 2100. The autoregressive integrated moving average (ARIMA) model (Box and Jenkins 1976) is one of the most widely used time series models (Han et al. 2010). The popularity of ARIMA model in many areas is due to its flexibility and the systematic searching at each stage (identification, estimation and diagnostic check) for an appropriate model. The ARIMA model approach has several advantages over others, such as moving average, exponential smoothing, neural network, and in particular, its forecasting capability and its richer information on time-related changes (Yurekli et al. 2005). In order to identify the appropriate ARIMA model, the Box-Jenkins method was used. Since the current research deals with climate data that follow a non-seasonal cycle, the non- seasonal ARIMA (p,d,q) was used (Hosking 1981). A non-seasonal ARIMA model is classified as an 'ARIMA (p,d,q)' model, where: p is the number of autoregressive terms d is the number of non-seasonal differences needed for stationarity q is the number of lagged forecast errors in the prediction equation

In the identification step data transformation is often required to make the time series stationary. Stationarity is a necessary condition in building an ARIMA model used for forecasting. The original time-series are non-stationary and they are all differenced at lag d = 1. The rest of the model's parameters determined by the autocorrelation function (ACF) and the partial autocorrelation function (PACF) plots and the fitting models are evaluated by examining the residuals with ACF, the Ljung-Box test (Ljung and Box 1978), and comparing the akaike information criteria (AIC) values of each model. Following the above mentioned steps, the appropriate ARIMA model was determined to be ARIMA (1,1,1). A sample was used for validation by holding out 30% of the data set (8 years) to check the validity of the aforementioned model ARIMA (1,1,1). Additionally, Relative Mean Square Error (RMSE) was estimated in order to quantify the performance of the model, whose equation (Eq. 3.09) has the following form:

59

n 2 (Xobs,i  X model,i ) RMSE  i1 n (3.09)

Where:

Xobs is observed values

Xmodel is modelled values at time/place i n is the sample size

3.12 Growing Degree Days

Climate affects practically all the physiological processes that determine plant life (IPCC 2014). A major challenge and objective in agriculture is to predict the occurrences of specific physical or biological events. For this reason, flower phenology has been widely used to study the flowering in plant species of economic interest, and in this concept, temperature and heat units have been widely accepted as the most important factors affecting processes leading to flowering. The determination of heat requirements in the first developing phases of plants has been expressed as Growing Degree Days (GDD). Determination of GDD is useful for achieving a better understanding of the flowering season development in several plant species, and for forecasting when flowering will occur (Orlandi et al. 2005). Temperature and GDD represent two important spatially-dynamic climatic variables, as they both play vital roles in influencing forest development by directly affecting plant functions such as evapotranspiration, photosynthesis, and plant transpiration. Understanding the variations and the spatial distribution of GDD is crucial to the sustainable agricultural and forest management, as GDD relates to the integration of growth and provides precise point estimates (Hasan et al. 2007; Matzarakis et al. 2007). The canonical form of the equation (Eq. 3.10) that is used to calculate the GDD is:

(3.10)

Where: o Tmax represents the daily maximum temperature ( C) o Tmin represents the daily minimum temperature ( C)

Tbase represents the temperature below which the process of growth does not progress

60

Tbase varies among species and cultivars (Wang 1960; McMaster and Wihelm 1997).

In some cases, in order to simplify the equation 3.10, the quantity (Tmax + Tmin) / 2 is set equal to Tavg, although this is not permissible and can lead to false estimations (Matzarakis et al. 2007). Many modifications have been suggested to enhance the biological meaning of Eq. 3.10, such as incorporating an upper temperature threshold (Gilmore and Rogers 1958; McMaster and Smika 1988), converting to photo thermal units by adding a photoperiod variable (Masle et al. 1989), using only the maximum or minimum temperature or portion of the day (Cross and Zuber 1972; Masle et al. 1989), and incorporating functions for other environmental factors that affect phenology or the process being considered (e.g. water, nutrients, light quality or quantity, CO2; McMaster et al. 1992; Wilhelm and McMaster, 1995). Additionally, Cross and Zuber (1972) and Perry et al. (1986) found that the Eq. 3.10 was used under many different forms and ways and the users were unaware of the significant errors that they could have emerged by these differences (Matzarakis et al. 2007). To address this problem, the two most common interpretations of Eq. 3.10 are mentioned to calculate GDD. In the first approach, (Tmax + Tmin) / 2 < Tbase, the first term of this inequality takes the form (Tmax + Tmin) / 2 < Tbase. The certain approach is the most widespread for calculating the GDD particularly in simulation models (Goyne et al. 1977; Gallagher 1979; Davidson and Campbell 1983; Narwal et al. 1986; Jefferies and Mackerron 1987; McMaster and Smika 1988; Mathan 1989; Masle et al. 1989; Kirby 1995) and it is mostly used by researchers and practitioners involved in small grain production.

In the second approach, the separate values of Tmax and Tmin are compared independently with Tbase. If Tmax < Tbase then Tmax = Tbase and if Tmin < Tbase, then Tmin = Tbase. The current approach is the most commonly used method in calculating GDD for corn, but it is also used for other crops as well (Tollenaar et al. 1979; Russelle et al. 1984; Baker et al. 1986; Edwardson and Watt 1987; Wilhelm et al. 1987, 1989; Bauer et al. 1988; Cutforth and Shaykewich 1989; Ketring and Wheless 1989; Masoni et al. 1990; Swanson and Wilhelm 1996). Data regarding the main cultivations in the study areas are collected by the Hellenic Statistical Authority2. Table 3.12 depicts these main cultivations and their characteristics in every study area. An attempt is performed through the literature to estimate the necessary GDD units that are required for every cultivations to reach its maturity in the study areas, although these are characterized by different climate conditions and thus the required GDD units for each cultivation may differ.

2 www.statistics.gr 61

Table 3.12. Main cultivations and their characteristics

Main GDD units Study area Latin name References cultivations to maturity Asparagus Asparagus officialis 530-660 Morrison et al. (2014) Sugar Beet Beta vulgaris 1400-1500 Miller et al. (2001) Gossypium Miley and Oosterhus Ardas River Cotton 1700-2100 hirsutum (1990) basin Maize Zea mays 1360-1630 Matzarakis et al. (2007) Sunflower Helianthus annuus 2200-2800 Raes et al. (2010) Cherry Prunus avium 800+ Miller et al. (2001)

Gossypium Miley and Oosterhus Cotton 1700-2100 hirsutum (1990) Wheat Triticum aestivum 1550-1680 Matzarakis et al. (2007) Sperchios Maize Zea mays 1360-1630 Matzarakis et al. (2007) River basin Olive Olea europea 900+ Miller et al. (2001) Allan and Ferguson Pistachio Pistachia vera 2000-2600 (2014)

Olive Olea Europea 900+ Miller et al. (2001) Geropotamos Grape vine Vitis vinifera 1210-1844 Köse (2014) River basin Solanum Tomato 1500-2000 Raes et al. (2010) lycopersium

62

Chapter 4

Results

4.1 Sensitivity analysis of PET formulae

Figures 4.01 - 4.05 depict the beanplots of the mean daily values of PET for all the examined formulae in the current study, as well as the 25th, 50th and 75th percentile of the values. Tables 4.01 - 4.05 present the results of the sensitivity analysis of the PET formulae for all the study areas. The results of the meteorological variables are measured in terms of percentage and the Potential Evapotranspiration values in mm.

Figure 4.01. Beanplots of the mean daily PET (mm) for Orestiada station in Ardas River basin, based on the PET formulae examined in Table 3.08

Figure 4.02. Beanplots of the mean daily PET (mm) for Metaxades station in Ardas River basin, based on the PET formulae examined in Table 3.08

63

Figure 4.03. Beanplots of the mean daily PET (mm) for Lamia station in Sperchios River basin, based on the PET formulae examined in Table 3.08

Figure 4.04. Beanplots of the mean daily PET (mm) for Tympaki station in Geropotamos River basin, based on the PET formulae examined in Table 3.08

Figure 4.05. Beanplots of the mean daily PET (mm) for Zaros station in Geropotamos River basin, based on the PET formulae examined in Table 3.08

64

According to the beanplots which constitute a widely used technique for descriptive statistics (Muthers and Matzarakis 2010), the highest daily PET values are presented in the Geropotamos River basin, which is located in the southest point of the 3 study areas and it is characterized purely by Mediterranean conditions (Csa) (Peel et al 2007). On the other hand, the PET values in the Ardas River basin which is characterized mainly by continental climate (Dfb), and partly by temperate (Cfb) and humid subtropical climate (Cfa), are lower compared to the other study areas. Furthermore, some minor differentiations are observed amongst the examined stations of the Ardas River basin and the other study areas. Specifically, according to the figures 4.01 and 4.02, in both stations of the Ardas River basin the values of PET are mainly concentrated between 1-3mm, while in the Sperchios and Geropotamos River basins the values present a higher concentration between 2-4mm. This is validated due to the higher temperatures that exist in the Sperchios and Geropotamos study areas (influenced by the Mediterranean climate), compared to the temperatures in the Ardas River basin that are slightly lower (Matzarakis 2006). Some individual differentiations also appeared firstly in the Ardas River basin for the

PETHar equation, which according to the beanplots presents a high concentration of values near 6-7mm, while in the other study areas the values for the certain equation are lower.

Secondly, for the 3 versions of Hamon, PETHam1, PETHam2, PETHam3 for the Sperchios and Geropotamos River basins the higher values present low concentration although their range in these two study areas is larger than the Ardas River basin. Thirdly, the PETRom equation overestimates PET in the Geropotamos River basin for about 200-400mm as opposed to the other study areas. The reason for this overestimation lies in the fact that PETRom is a temperature-based equation which, during the sensitivity analysis showed that it is influenced by more than 50% from Tmin. According to the preliminary data analysis, Tmin appeared with great difference that exceeds 4oC between the northern (Ardas River basin) and the southern

(Geropotamos River basin) part of Greece. Lastly, PETASCE in 'Tympaki' station (from 'Zaros' station there was no available wind speed data series) is also appeared with differentiations compared to the other study areas. The values present high concentration between 1-2mm and this fact makes this examined equation an unreliable one for the Geropotamos River basin. For the simplicity and better understanding, all results of the sensitivity analysis were converted into percentages as depicted in Tables 4.01 - 4.05.

65

Table 4.01. Sensitivity analysis results (%) - Orestiada Table 4.02. Sensitivity analysis results (%) - Metaxades

Ardas River basin - Orestiada station Ardas River basin - Metaxades station

Wind PET Eq. T T T T RH Wind speed DL R R R PET(mm) PET Eq. T T T T RH DL R R R PET(mm) mean min max dew s a n mean min max dew speed s a n PET 26.6 28.8 43.6 1.0 1131.7 PET 19.2 31.8 48.0 1.0 1048.7 Har Har PET <0.01 <0.01 99.9 <0.01 2336.3 PET <0.01 <0.01 99.9 <0.01 2246.0 PT PT PET 1.1 13.4 34.2 51.3 1195.7 PET 1.1 15.2 32.3 51.4 1195.6 Ham1 Ham1 PET 99.2 0.8 1243.5 PET 97.8 2.2 1282.1 Ham2 Ham2 PET 5.1 36.7 57.4 0.8 844.2 PET 4.5 35.7 58.6 1.2 844.3 Ham3 Ham3 PET 97.8 2.2 1291.1 PET 95.9 4.1 1318.9 McG McG PET 5.7 51.2 34.5 8.6 1609.5 PET 4.2 51.4 35.2 9.2 1503.2 Rom Rom PET 64.8 35.2 1121.0 PET 52.8 47.2 1051.1 Cap Cap PET 61.4 38.6 1186.8 PET 49.7 50.3 1110.0 Jen Jen PET 52.7 47.3 1044.1 PET 41.8 58.2 955.0 Tur Tur PET 4.9 95.1 804.7 PET 3.5 96.5 751.6 Iva Iva PET <0.01 2.5 2.3 <0.01 <0.01 95.2 1176.7 PET <0.01 2.1 3 <0.01 <0.01 94.9 1232.8 ASCE ASCE 1255.3 1209.7

Sperchios River basin - Lamia station Geropotamos River basin - Tympaki station

Wind PET Eq. T T T T RH Wind speed DL R R R PET(mm) PET Eq. T T T T RH DL R R R PET(mm) mean min max dew s a n mean min max dew speed s a n PET 22.2 30.6 45.3 1.9 1086.9 PET 21.3 31.0 46.8 0.9 1185.8 Har Har PET 1.1 38.7 59.1 1.1 2297.1 PET < 0.01 99.3 0.7 < 0.01 2465.0 PT PT PET 1.2 17.0 31.3 50.5 1221.2 PET 0.1 18.4 30.9 50.5 1271.1 Ham1 Ham1 PET 98.4 1.6 1272.3 PET 99.3 0.7 1308.1 Ham2 Ham2 PET 5.6 37.3 55.4 1.7 859.8 PET 5.7 38.4 55.1 0.8 893.5 Ham3 Ham3 PET 97.7 2.3 1391.8 PET 97.5 2.5 1510.8 McG McG PET 4.6 53.4 34.3 7.7 1703.4 PET 4.3 56.4 31.9 7.4 1874.5 Rom Rom PET 61.2 38.8 1070.2 PET 59.1 40.9 1160.6 Cap Cap PET 58.2 41.8 1128.9 PET 56.4 43.6 1220.9 Jen Jen PET 47.1 52.9 1005.0 PET 43.2 56.8 1145.7 Tur Tur PET 4.5 95.5 851.7 PET 4.5 95.5 937.2 Iva Iva

PETASCE < 0.01 2.4 2.6 < 0.01 < 0.01 95 1466.7 PETASCE < 0.01 3.4 1.6 < 0.01 < 0.01 95 908.6

1262.6 1361.2

Table 4.03. Sensitivity analysis results (%) - Lamia Table 4.04. Sensitivity analysis results (%) - Tympaki 66

Table 4.05. Sensitivity analysis results (%) - Zaros station

Geropotamos River basin - Zaros station

PET Eq. Tmean Tmin Tmax Tdew RH Wind speed DL Rs Ra Rn PET(mm)

PETHar 21.7 30.9 46.6 0.8 1154.1

PETPT < 0.01 79.6 20.4 < 0.01 2414.9 PETHam1 1.3 17.8 31.2 49.7 1247.8

PETHam2 99.3 0.7 1291.2

PETHam3 5.5 37.8 56.0 0.7 878.8

PETMcG 97.7 2.3 1480.8

PETRom 3.2 58.9 30.2 7.7 2075.0

PETCap 60.1 39.9 1139.6 PETJen 57.3 42.7 1200.7

PETTur 45.1 54.9 1188.9

PETIva 5.2 94.8 1037.5

PETASCE n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1373.6

In Ardas River basin, the PET equation that describes more satisfactorily the PET conditions prevailing in the area and is closer to the average results is the PETHam2 equation. Noteworthy is the fact that there is a differentiation between 'Orestiada' and 'Metaxades' stations for the Ardas River basin, as in the 'Metaxades' station the inter-comparison of the PET equations indicated that the formula that estimates PET with the highest accuracy is the

PETHam1 equation. Regarding the equations that estimated the PET values in the Ardas River basin with great efficiency, PETJen is the only formula in which the influence of the contributing variables is almost equal (Tmean and Rs). In all the other equations PETHam1, PETMcG and

PETASCE one of the variables in every equation (In the case of PETHam1, PETMcG the Tmean and in

PETASCE the Rn) is influencing the results almost entirely. Specifically, during PETJen sensitivity analysis for the 'Metaxades' station, the results were almost similar for both variables (Tmean =

49.7% and Rs = 50.3%) which indicates that the equation is able to estimate the results very satisfactorily, while taking into account both variables equally.

During the sensitivity analysis of the PET formulae for the meteorological station of 'Lamia' in the Sperchios River basin, the results indicated that the most suitable PET equations are PETHam2 and PETHam1. These equations are temperature-based as PETHam2 uses Tmean and

Day Length, while PETHam1 uses Tmin, Tmean, Tmax and Day Length as input data, respectively.

Especially for the PETHam2 equation, the results were almost identical to the average PET results of all the formulae. In the Geropotamos River basin, the results from the application of the PET formulae were very similar to those in the Sperchios River basin. PETHam2 and PETHam1 can estimate PET

67 adequately, although PETJen in 'Tympaki' station which is located near the coastline in a low altitude, can also be used to estimate PET. Table 4.06 presents the ranking of the PET formulae, according to the average results. For better representation and understanding of the ranking, each formula was given a colour. Since wind speed data weren't available for the 'Zaros' station in the Geropotamos River basin,

PETASCE equation couldn't be applied and it was given a grey colour.

Table 4.06. Ranking of PET formulae according to the average results of each station

Ardas River basin Sperchios River basin Geropotamos River basin

Orestiada Metaxades Lamia Tympaki Zaros PET Eq. PET Eq. PET Eq. PET Eq. PET Eq.

PETHam2 PETHam1 PETHam2 PETHam2 PETHam2

PETMcG PETASCE PETHam1 PETHam1 PETMcG

PETHam1 PETHam2 PETMcG PETJen PETHam1

PETJen PETJen PETJen PETMcG PETJen

PETASCE PETMcG PETHar PETHar PETTur

PETHar PETCap PETCap PETCap PETHar

PETCap PETHar PETASCE PETTur PETCap

PETTur PETTur PETTur PETIva PETIva

PETRom PETRom PETHam3 PETASCE PETHam3

PETHam3 PETHam3 PETIva PETHam3 PETRom

PETIva PETIva PETRom PETRom PETPT

PETPT PETPT PETPT PETPT PETASCE

Regarding the differentiations that were emerged in every PET formula for each study area, it is of great interest to mention and analyse first of all the differences at the 'Metaxades' station during the sensitivity analysis and inter-comparison of PETCap, PETJen and PETTur. All three mentioned equations are radiation-based equations and during the sensitivity analysis showed that they are influenced almost equally by air temperature and solar radiation.

Nevertheless, except from PETJen which, according to table 4.06 is very high in the rankings, the other 2 equations cannot be used to estimate satisfactorily the PET conditions prevailing in the selected study areas.

PETPT is presented with great heterogeneity in the 3 study areas regarding the sensitivity analysis and the results of PET present great differentiations compared to the average results. It constitutes a method that is not suitable for the certain study areas because it greatly overestimates the results.

68

According to table 4.06, PETHam2 can be used in Sperchios and Geropotamos River basins. Furthermore, these study areas face the same climatic conditions. Thus, in the ranking table they do not present great differentiations and sensitivity analysis produced similar results regarding to which PET method should be used in areas that face Mediterranean conditions. On the other hand, in the Ardas River basin, the inter-comparison is appeared with differentiations regarding to which PET formula is the most appropriate to estimate PET. Although the stations of 'Orestiada' and 'Metaxades' are located very close to each other (distance is approximately 27 km), the different climate conditions that appeared in the certain study area operate as an inhibitor towards a safe evaluation of the most appropriate PET method. Nonetheless, PETHam2 can be used for the estimation of potential evapotranspiration in the Ardas River basin.

PETHam2 and PETHam1 equations that according to the results of the current research can be used for the estimation of PET in all the current study areas are temperature-based equations. During the estimation of PET using the PETHam2 method, the results are almost exclusively affected by Tmean, while applying PETHam1 the results are equally affected by air temperature (Tmax, Tmin and Tmean, according to the order in which they affect) and Day Length (DL) as well.

PETIva equation is the only method that apart from air temperature, uses relative humidity (RH) input data as well. The results indicated that there is a relatively large error between the estimation of the method results and the mean annual results. Additionally, the

PETIva equation is greatly influenced by relative humidity values (lowest percentage is observed in 'Zaros' station in the Geropotamos River basin and it is equal to 94.8%, while for the

'Metaxades' station the influence rate of relative humidity in the PETIva equation is 96.5%).

Noteworthy is the fact that although PETASCE is considered the most representative equation for the estimation of PET and it is widely used in many studies, during its sensitivity analysis, the equation was found to be influenced very strongly (almost 95%) in every case, by

Rn. On the other hand, Tmean, wind speed and Tdew do not have any significant influence on the results of the equation (values < 0.01%). Nevertheless, when PETASCE was applied, the results were very similar to the average values and further indicate that PETASCE can also be used in every case for the estimation of PET. Additionally, PETMcG according to table 4.06 is ranked as the 2nd most satisfactorily equation in 'Orestiada' station in Ardas River basin and 'Zaros' station in Geropotamos River basin.

69

4.2 Integrated analysis of Precipitation

Tables 4.07, 4.08 and 4.09 present the results of the integrated analysis of present and future precipitation for the selected study areas of the current study. Figures 4.06, 4.07 and 4.08 depict the mapping of the precipitation for the present, as well as for 2021-2050 and 2071-2100 periods under A1B and B1 scenarios, respectively.

Table 4.07. Precipitation analysis results of present and future precipitation conditions in the study areas

Ardas River basin - Precipitation (mm) Scenario/Period A1B B1 A1B B1 Current 839.8 2021-2050 499.1 555.3 -40.6% -33.9% 2071-2100 450.8 468.6 -46.3% -44.2%

Sperchios River basin - Precipitation (mm) Scenario/Period A1B B1 A1B B1 Current 792.9 2021-2050 550.5 582.3 -30.6% -26.6% 2071-2100 475.6 538.5 -40.0% -32.1%

Geropotamos River basin - Precipitation (mm) Scenario/Period A1B B1 A1B B1 Current 759.8 2021-2050 418.6 427.2 -44.9% -43.8% 2071-2100 372.8 391.5 -50.9% -48.5%

Integrated analysis of precipitation for the Ardas River basin indicated that precipitation is expected to be reduced for both applied IPCC future scenarios, A1B and B1. The values mentioned in table 4.07 correspond to the average annual values over the whole domain. In the 1st case for the A1B scenario, precipitation is expected to be reduced by 340.7mm (-40.6%) for the period 2021-2050 and this reduction will reach the 388.9mm (-46.3%) for the period 2071- 2100. On the other hand, the B1 scenario is more optimistic, although it is also presented with a reduction, in the 1st period of 2021-2050 by 2021-2050 by 284.5 (-33.9%) and in the 2nd period of 2071-2100 by 371.2mm (-44.2%).

70

Regarding the integrated analysis of future precipitation responses in the Sperchios River basin, the results indicated that future precipitation is expected to be reduced as well by 242.4mm (-30.6%) for the period 2021-2050 and by 371.3mm for the period 2071-2100, under the A1B scenario. Especially for the 2nd examined period of 2071-2100, the results indicated that the amount of precipitation is expected to be reduced by 40% compared with the reference periods' data. For the B1 scenario, the results appeared with more moderate values, but also reduced by 210.6mm or -26.6% (2021-2050), and 254.4mm or -32.1% (2071-2100), respectively. Concerning the Geropotamos River basin, the current study area is presented with the greatest differentiations, compared with the other examined study areas. Namely, the results present a mutual agreement and conclude that precipitation is expected to be reduced by 341.2mm, which is about 45% of the current precipitation, reduction that will reach almost 387mm, or 50% of the current values, according to A1B scenario which is presented as the most pessimistic between the examined scenarios.

Table 4.08. Trend analysis results of present and future precipitation conditions in the study areas

Study area Ardas Sperchios Geropotamos Scenario A1B B1 A1B B1 A1B B1 Z value -2.05 -1.13 -2.07 -1.06 -2.62 -2.29

Respecting the total trend analysis of precipitation using the Mann-Kendall test that is depicted in table 4.08, Z values for both applied scenarios showed a downward trend (negative values) with -2.05 for A1B and -1.13 for the B1 scenario, respectively. On the Sperchios River basin, the Z values will also follow a downward trend, with -2.07 for A1B scenario and -1.06 for the B1 scenario, while the application of the Mann-Kendall trend test for the Geropotamos River basin indicated that a strong downward trend of precipitation values with A1B Z value = -2.62, and B1 Z value = -2.29 will be presented, which constitutes the highest amongst the examined study areas.

71

Table 4.09. Seasonal analysis results of present and future precipitation conditions in the study areas

Ardas River basin - Precipitation (mm) Winter 287.8 Spring 182.8 Current Summer 150.3 Autumn 218.9 Scenario/Period A1B B1 A1B B1 mm % Winter 170.5 169.3 -40.8% -41.2% Spring 119.7 132.6 -34.5% -27.4% 2021-2050 Summer 108.4 109.4 -27.9% -27.2% Autumn 100.5 144.0 -54.1% -34.2% Winter 158.9 148.0 -44.8% -48.6% Spring 103.8 109.7 -43.2% -40.0% 2071-2100 Summer 111.6 94.6 -25.7% -37.1% Autumn 76.5 116.4 -65.1% -46.8%

Sperchios River basin - Precipitation (mm) Winter 237.4 Spring 241.2 Current Summer 68.9 Autumn 245.4 Scenario/Period A1B B1 A1B B1 mm % Winter 193.5 207.4 -18.5% -12.6% Spring 145.9 151.5 -39.5% -37.2% 2021-2050 Summer 47.2 48.7 -31.5% -29.3% Autumn 163.9 174.7 -33.2% -28.8% Winter 153.9 182.5 -35.2% -23.1% Spring 113.9 137.8 -52.8% -42.9% 2071-2100 Summer 45.2 46.2 -34.3% -32.9% Autumn 162.5 172.0 -33.8% -29.9%

Geropotamos River basin - Precipitation (mm) Winter 254.5 Spring 257.4 Current Summer 18.8 Autumn 229.0 Scenario/Period A1B B1 A1B B1 mm % Winter 165.6 159.0 -34.9% -37.5% Spring 120.5 136.2 -53.2% -47.1% 2021-2050 Summer 12.7 10.7 -32.1% -42.8% Autumn 119.8 121.3 -47.7% -47.0% Winter 154.2 150.8 -39.4% -40.8% Spring 105.9 124.7 -58.9% -51.6% 2071-2100 Summer 12.8 6.6 -32.0% -64.7% Autumn 99.9 109.4 -56.4% -52.2%

72

During the seasonal analysis of the future response of precipitation for the Ardas River basin, all seasons are presented with reduction which is very strong for the winter and autumn, and more mild for spring and summer. Specifically, in both applied scenarios, winter season precipitation, under the A1B scenario is expected to be reduced by -40.8% during 2021-2050 and it can reach up to -44.8% for the 2071-2100 period. Under the B1 scenario, a reduction of - 41.2% for the years 2021-2050 will be followed by a further reduction of -48.6% for the period 2071-2100. Summer season is presented with very small differentiations regarding the future response of precipitation over Ardas River basin. Specifically, A1B scenario showed a reduction of -27.9% (2021-2050) and -25.7% (2071-2100), respectively. B1 scenario though, showed some differentiations during the simulation periods of precipitation that in worst case scenario (B1: 2071-2100) could reach 55.7mm. A noteworthy fact is that winter is the only season that B1 scenario simulation values for both future periods appeared more decreased than A1B values. Seasonal analysis of future precipitation in the Sperchios River basin resulted that during the application of A1B scenario, spring season will face the greatest reduction by -39.5% for the examined period of 2021-2050, percentage that will reach -52.8% for the examined period of 2071-2100. Additionally, according to B1 scenario, spring season will also face the greatest reduction during the year, compared with the other seasons. Generally, although the future precipitation Sperchios River basin will be decreased as occurred by the analysis of the results, it will face the least reductions, compared with the other study areas. Respecting the seasonal analysis of precipitation for the Geropotamos River basin, all seasons are presented with high reduction in precipitation for every scenario and every examined future period.

73

Figure 4.06. Precipitation analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

74

Figure 4.07. Precipitation analysis of the Sperchios River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

75

Figure 4.08. Precipitation analysis of the Geropotamos River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

76

During the spatial interpolation of the present and future precipitation in the Ardas River basin it is evident that in the worst case scenario (B1: 2071-2100) annual precipitation in some cases is expected to be less than 400mm in the Ardas valley, and it will not exceed 600mm in the biggest part of the mountainous areas. Nonetheless, for the 1st examined period of B1 scenario, annual precipitation levels will reach almost 800mm in a considerable part of the basin. Spatial interpolation of precipitation in the Sperchios River basin indicated that according to the worst-case scenario (A1B: 2071-2100), precipitation will be reduced by more than 600mm and it will not exceed 400mm. Nevertheless, according to B1 scenario for the period 2021-2050, a sizable part of the Sperchios River basin will still receive up to 800mm in the near future, amount very crucial for the agriculture production. For the 1st examined period of 2021-2050 in the Geropotamos River basin, during the spatial interpolation of precipitation both applied scenarios resulted almost similar values that range from 0-400mm in the Messara valley and up to 800mm in a very small area located in the North side near mount Psiloreitis. During the simulation of the 2nd future period for the years 2071-2100, spatial distribution of both scenarios is also presented with similarities but at this case, precipitation values will be further reduced and will not exceed 600mm.

4.3 Present and future responses of runoff

All the necessary data were given as inputs into the model and several simulations were run for the reference, 2021-2050, 2071-2100 periods, for A1B & B1 scenarios in every study area. Figure 4.09 gives a schematic representation between the observed and the simulated discharge for the calibrated (reference) period for the study areas, respectively. Table 4.10 presents the results of the simulations for the study areas as well as the coefficient of efficiency (EF) values. The major contributors -apart from runoff- of the hydrological cycle i.e. Actual Evapotranspiration (AET), was resulted as an output through the model simulation, while mean annual precipitation (PREC) values for the same reference as well as future periods were given as inputs from the results of the integrated precipitation analysis. In order to depict the complete concept of the procedures that are taking place in the hydrological cycle, table 4.10 was enriched with information regarding the water that it is stored in the soil (SW) as well as ΔS which represents the change in storage which means a change in the water volume in any number of 'buckets' in ArcSWAT namely: shallow aquifer, deep aquifer, soil moisture, and impoundments when exist (reservoirs, ponds, wetlands, and potholes). The volume in these 'buckets' can go up or down depending on input versus output, which changes over the model runs. Moreover, regarding the ΔS as it was resulted as an output from the model, it is referred in

77 the amount of water (mm) that occurred by the interactions amongst the 'buckets' by the end of a specific analysis period. Figure 4.10 and table 4.11 depict the trends analysis results for the years 2000-2100 divided in 10-year intervals that also display max and min values for better representation and understanding of the results. Table 4.12 presents the average irrigation water withdrawals and the agricultural water withdrawals versus total water withdrawals for Greece during 1988-2012 as well as the forecasted outputs from the ARIMA model for the periods 2021-2050 and 2071- 2100.

Figure 4.09. Observed and simulated discharge (m3/s) for examined reference periods in the study areas

78

Table 4.10. ArcSWAT simulation results (the brackets depict the difference between the reference period's and the future periods' values in terms of percentage)

Ardas River basin Discharge Q (Observed) Discharge Q (Simulated) PREC AET (mm) SW (mm) ΔS (mm) (m3/s) (m3/s) (mm) (mm) (mm) (mm) (mm) Reference period 54.7 51.3 223.5 839.8 420.1 166.8 29.5 A1B 48.0 162.6 (-6.4%) 499.1 (-40.6%) 190.3 (-54.7%) 122.8 (-26.4%) -23.1 2021-2050 B1 50.4 175.7 (-1.8%) 555.3 (-33.9%) 193.6 (-53.9%) 130.2 (-21.9%) 12.0 A1B 40.3 120.4 (-21.5%) 450.8 (-46.3%) 194.5 (-53.7%) 110.7 (-33.6%) -29.8 2071-2100 B1 42.6 133.0 (-16.9%) 468.6 (-44.2%) 197.5(-53.0%) 106.9 (-35.9%) -21.4 EF = 0.93

Sperchios River basin Discharge Q (Observed) Discharge Q (Simulated) PREC AET (mm) SW (mm) ΔS (mm) (m3/s) (m3/s) (mm) (mm) (mm) (mm) (mm) Reference period 15.5 17.6 250.1 792.9 424.5 127.4 -9.1 A1B 14.3 203.2 (-18.6%) 550.5 (-30.6%) 251.5 (-40.8%) 91.3 (-28.3%) 4.5 2021-2050 B1 15.0 213.1 (-14.7%) 582.3 (-26.6%) 260.1 (-38.7%) 102.5 (-19.5%) 6.6 A1B 12.0 170.5 (-32.0%) 475.6 (-40.0%) 250.7 (-40.9%) 74.2 (-41.8%) -19.8 2071-2100 B1 13.8 196.1 (-21.4%) 538.5 (-32.1%) 268.9 (-36.7%) 90.7 (-28.8%) -17.2 EF = 0.91

Geropotamos River basin Discharge Q (Observed) Discharge Q (Simulated) PREC AET (mm) SW (mm) ΔS (mm) (m3/s) (m3/s) (mm) (mm) (mm) (mm) (mm) Reference period 10.6 12.9 176.7 759.8 421.6 136.6 24.9 A1B 12.0 164.3 (-7.1%) 418.6 (-44.9%) 191.0 (-54.7%) 77.9 (-43.0%) -14.6 2021-2050 B1 11.2 153.4 (-11.2%) 427.2 (-43.8%) 190.8 (-54.8%) 84.5 (-38.1%) -1.5 A1B 10.6 145.2 (-18.3%) 372.8(-50.9%) 196.5 (-53.4%) 67.5 (-50.6%) -36.4 2071-2100 B1 11.0 150.7 (-14.5%) 391.5 (-48.5%) 195.5 (-53.6%) 69.8 (-48.9%) -24.5 EF = 0.64

Note: PREC=Precipitation (mm); AET=Actual Evapotranspiration (mm); SW=Amount of water stored in soil profile in watershed (mm); ΔS=change in storage (mm) 79

During the calibration process of the model, special emphasis was given on the successful representation of the annual discharge. The main initial objective of the calibration procedure was the reproduction of realistic and acceptable values of the observed streamflow data series. As a first remark, the EF coefficient shows great effectiveness for the examined study areas and especially for the Ardas and the Sperchios River basins. These values were very close to the optimal value of 1.0, and imply that the model was able to simulate the streamflow almost with the same efficiency as the observed values. Figure 4.09 depicts the observed and the simulated streamflow (m3/s) for the study areas. ArcSWAT was able to simulate with great efficiency (EF = 0.93) the streamflow procedure in the Ardas River basin, while in the Sperchios River basin EF was equal to 0.91, and 0.64 in the Geropotamos River basin, respectively. Regarding the streamflow, for the simulated period of 2021-2050, it is expected to be reduced in every study area for both applied scenarios. The greatest reduction will be observed in the Sperchios River basin, followed by the Geropotamos and the Ardas River basins, accordingly. For the period of 2071-2100, streamflow is expected to further decrease, and in the Sperchios River basin this reduction will reach up to -32% of the initial simulated values for the A1B scenario. During the 2nd calibration period of 2071-2100, the Ardas River basin is appeared also with critical reductions that will reach up to -21.5%, while in the Geropotamos River basin the streamflow will be reduced by more than -18%. Regarding the Actual Evapotranspiration that was also exported from the simulations of the model, for the first period of 2021-2050, it is expected to decrease in all the study areas compared with the reference period analysis. Nevertheless, during the years 2071- 2100, the AET is expected to increase for both scenarios.

80

Figure 4.10. Schematic representation of trend analysis results regarding the variation of runoff in the upcoming years in the study areas

Table 4.11. Trend analysis results of present and future runoff conditions in the study areas

Study area Ardas Sperchios Geropotamos Scenario A1B B1 A1B B1 A1B B1 Z value -1.68 -1.61 -1.15 -0.98 -0.65 -0.61

Trend analysis indicated that the Ardas River basin will face the strongest downward trend, followed by the Sperchios and the Geropotamos River basins, respectively. Moreover, the Ardas River basin is appeared with the greatest variations which are very abrupt in many cases for both scenarios, while the variations in the Geropotamos and the Sperchios River basins will be smoother in the upcoming years, as it was depicted in Figure 4.10.

81

Table 4.12. Average irrigation water withdrawal in Greece and agricultural water withdrawal versus total water withdrawal

Irrigation water withdrawal Agricultural water withdrawal (109 m3/year) Vs total water withdrawal (%) 1988 -2012 8.0 90.1 2021-2050 8.8 73.9 2071-2100 10.0 50.1

In order to further understand how these variations of the aspects of the hydrological cycle and specifically streamflow will impact the following years the sustainable development of the environment as well as the society, data regarding the irrigated water withdrawal were collected by FAO for the period of 1988-2012, and using the ARIMA (p, d, q) model were forecasted up to 2100. Initially, the RMSE was estimated for the irrigation water withdrawal

(RMSEiww) and the agricultural water withdrawal (RMSEaww). The RMSEiww = 0.68, while the

RMSEaww = 13.05, which indicates that for the years that the validation was performed, the model was able to perform a satisfactory prediction regarding the comparison of observed and simulated values. To better understand the future variations of the aspects of the hydrological cycle, table 4.13 was created, based on the percentage changes of table 4.10, presenting the study areas, the variables of the hydrological cycle and their classification, regarding the severity of the future climate change that will prevail amongst these study areas. The numbers represent the classification having 1 as the most severe, 2 as moderate severe, and 3 as the least severe regarding to future climate change.

Table 4.13. Classification of the study areas regarding the severity of future climate change

Study area Discharge PREC AET SW Total Ardas (North) 3 2 2 3 3 Sperchios (Central) 1 3 3 2 2 Geropotamos (South) 2 1 1 1 1 Note: 1: most severe; 2: moderate severe; 3: least severe

From the results of table 4.13 it is evident that areas located in lower altitudes will be more vulnerable to future climate change regarding the variations of streamflow in comparison with the other aspects of the hydrological cycle than areas located in higher altitudes. 82

4.4 Integrated analysis of Growing Degree Days

Table 4.14 depicts the validation results between the reference and the future periods' data. Tables 4.15 and 4.16 present the results of the integrated analysis of present and future GDD. The results were performed for an annual scale, as well as for three different periods within the year starting from April, since the goal of the current study was to imprint the GDD for the warm season in Greece. Figures 4.11, 4.12 and 4.13 depict the mapping of GDD for the present, 2021-2050 and 2071-2100 periods for the months from April to October under A1B and B1 scenarios, respectively.

Table 4.14. Reference (observed) and future (RCMs) simulations periods' data validation

Ardas Sperchios Geropotamos Tmax A1B B1 A1B B1 A1B B1 EF 0.86 0.81 0.80 0.85 0.90 0.91 std - (reference period) (mm) 0.7 0.3 0.4 std (2001-2100) (mm) 0.8 0.3 0.6 0.4 0.3 0.5 Difference (mm) 0.1 -0.4 0.3 0.1 -0.1 0.1

Ardas Sperchios Geropotamos Tmin A1B B1 A1B B1 A1B B1 EF 0.85 0.79 0.70 0.76 0.83 0.88 std - (reference period) (mm) 0.7 0.2 0.4 std (2001-2100) (mm) 0.6 0.2 0.4 0.4 0.2 0.3 Difference (mm) -0.1 -0.5 0.2 0.2 -0.2 -0.1 Note: std is the standard deviation

As a first remark, the coefficient of Efficiency shows great efficient for all the study areas as the values were very close to the optimal value of 1.0 and imply also in this certain case that the RCMs output simulation values that were adopted and used in the current study for maximum and minimum air temperature can describe with great efficiency the observed values. This is also validated by the standard deviation values which they do not present significant variations.

83

Table 4.15. Accumulated mean yearly GDD (oC) for reference, 2021-2050 and 2071-2100 periods

GDD 2021-2050 2071-2100 Study area Current A1B B1 A1B B1 Ardas River basin 1322.4 2022.7 2088.4 2241.0 2120.6 Sperchios River basin 1816.2 2506.6 2430.8 3043.2 2782.7 Geropotamos River basin 2763.8 2949.0 2915.9 3436.1 3225.0

Table 4.16. Accumulated mean GDD (oC) for reference (April-October), 2021-2050 and 2071-2100 periods

2021-2050 GDD Current Study area Apr-Sep Apr-Oct Apr-Nov (Apr-Oct) A1B B1 A1B B1 A1B B1 Ardas River basin 1102.6 1775.5 1648.8 1934.4 1817.9 1991.3 1865.7 Sperchios River basin 1783.4 2061.5 2012.0 2248.3 2205.1 2335.3 2287.7 Geropotamos River basin 2496.6 2146.9 2139.9 2268.9 2260.4 2487.7 2478.7

2071-2100 GDD Current Study area Apr-Sep Apr-Oct Apr-Nov (Apr-Oct) A1B B1 A1B B1 A1B B1 Ardas River basin 1102.6 1870.9 1670.3 2041.3 1996.7 2127.6 2062.7 Sperchios River basin 1783.4 2179.9 2196.7 2693.8 2490.2 2798.2 2583.3 Geropotamos River basin 2496.6 2230.5 2226.3 2703.0 2576.4 2856.8 2708.0

The analysis of the GDD in the study areas indicated that the growing degree units are expected to increase for both applied scenarios in every study area. Specifically, in the Ardas River basin according to table 4.15 the mean annual GDD will greatly increase in both examined periods and scenarios and will reach more than 2000 units. Additionally, for the examined sub-periods during the year a comparison between the GDD requirements for the main cultivations and the results showed that the future conditions will be more agreeable in the upcoming years. Regarding the Sperchios River basin, the Growing Degree Days units are expected to increase critically and they will reach up to 3000 units according to A1B scenario for the 2071-2100 period. During the sub-period analysis and specifically for the April-October period which constitutes the main period for agriculture activities in Greece, the GDD will exceed the 2200 units by 2021-2050, and will reach up to 2500-2600 units by the end of the

84 century. This implies that with the exception of the Pistachio cultivation that requires 2000- 2600 units, all the rest cultivations can find highly favourable conditions. Concerning the Geropotamos River basin, the present and the future analysis of GDD is presented with differentiations. Specifically, in the 1st examined period of 2021-2050 the GDD are expected to be reduced. Nevertheless, this is observed only during the sub-periods' analysis and not during the mean-inter-annual analysis. The reduction of the Growing Degree Days units in the Geropotamos River basin does not necessarily mean that that the existing cultivations will face problems in the future but in addition to the other study areas, the main cultivations in the Geropotamos River basin will not have the opportunity until the mid-century for further expansion and their extent will stay in the present levels. From this point and forward, the analysis illustrated that for the 2071-2100 period, the sub-period of April-October will present significant differentiations with very high GDD that will reach more than 2700 units, under the A1B scenario. The other examined sub-periods will not display great differentiations, although their GDD will increase.

85

Figure 4.11. Growing Degree Days analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

86

Figure 4.12. Growing Degree Days analysis of the Sperchios River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

87

Figure 4.13. Growing Degree Days analysis of the Geropotamos River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

88

The results of the spatial distribution of the GDD in the Ardas River basin indicated that in the upcoming years, the Ardas valley were the all agriculture productivity takes place will be a suitable place all the existing cultivations in the study area. Additionally, the favourable conditions will be expanded also in the highlands of the basin, with an exception on the northern part, were the Rhodope Mount is located and the climate classification becomes humid continental (Dfb). Concerning the spatial analysis for the Sperchios River basin, a prolonged part of the basin in the south side which according to the reference period's analysis is facing very low GDD values, will continue facing not favourable conditions for any cultivations. Nevertheless, almost the entire Sperchios valley is expected to reach more than 2500 units in the future, which constitutes the conditions on the certain study area very favourable for the existing cultivations. The future distribution of the GDD in the Geropotamos River basin will initially face some minor reductions and in the 'Messara' valley where the main agricultural production takes place, the conditions by the period of 2021-2050 will not be very favourable for all the existing cultivations. Nevertheless, by the end of the century, the GDD are expected to increase and the biggest part of the valley will be dominated by more than 2500 GDD units, and thus favourable conditions for all the existing cultivations.

4.5 Integrated analysis of Aridity

Table 4.17 depicts the average results for the total Aridity Index (AI) estimation, seasonal analysis and trend for the examined study areas. Subsequently, figures 4.14, 4.15 and 4.16 depict the spatial interpolation of the aridity index analysis (total analysis, seasonal analysis and trend) for the selected study areas.

Table 4.17. Aridity index analysis results for the selected Greek areas

Study area AI Analysis Winter Spring Summer Autumn Trend (Z values)

Ardas 0.94 1.31 0.68 0.25 1.05 -0.97 Sperchios 0.93 1.80 0.65 0.24 2.11 0.04 Geropotamos 1.09 1.72 0.41 0.14 1.61 -0.64 Note: mean values

As shown in table 4.17, the total Aridity Index analysis indicated that all the examined study areas face humid conditions during the year. The complex topography with the

89 presence of prolonged mountainous areas with high and moderate altitudes validates the results. Total Aridity Index analysis for the Geropotamos River basin resulted an AI = 1.09, which is the highest among the examined areas, although it is located in the southern part of the three study areas and constitutes a part of Crete Island which, through the examined literature above is an Island that faces drought conditions. Seasonal Aridity Index analysis indicated that winter is the most humid season, followed by autumn, spring and summer, respectively. The winter and autumn seasons face humid conditions, while spring faces semi-arid conditions. Summer season faces arid conditions with an AI = 0.14. The total analysis of the AI also indicated that the trend results follow a downward trend. Concerning the Ardas River basin, the total analysis of the Aridity Index indicated that the conditions are humid with an AI = 0.94. Seasonal analysis indicated that winter and autumn seasons face humid conditions, spring sub-humid and summer semi-arid conditions. At this case, the trend analysis showed a downward trend (Z value = -0.96), which is considered very significant as, according to the analysis, the aridity conditions tend to increase. Respecting the Sperchios River basin, the total Aridity Index is on the same level as the Ardas River basin, with AI = 0.93. Implementing the Mann-Kendal test, a Z value of 0.04 was estimated, which shows a minor upward trend of the aridity conditions prevailing in the Sperchios River basin.

90

Figure 4.14. Aridity Index analysis maps for Ardas River basin a)Total Analysis b) Spring c) Summer d) Autumn e) Winter

91

Figure 4.15. Aridity Index analysis maps for Sperchios River basin a)Total Analysis b) Spring c) Summer d) Autumn e) Winter

92

Figure 4.16. Aridity Index analysis maps for Geropotamos River basin a)Total Analysis b) Spring c) Summer d) Autumn e) Winter

93

Figure 4.14 depicts the aridity conditions analysis for the Ardas River basin. The analysis indicated that during the spatial interpolation of the AI, the mountainous areas present humid conditions, while the areas with moderate altitude display drier conditions during the year. These areas are located mainly in the Greek part of the study area, where all the agriculture activity takes place. Respecting the Sperchios River basin, the results of the AI analysis indicated that the aridity conditions prevailing in the study area range from very humid in the upper areas of the basin, to arid near the coastline. In the areas where the main part of Sperchios valley lies, the conditions are semi-arid. Seasonal AI mapping showed that during the spring and summer seasons the biggest part of the Sperchios valley faces semi-arid and arid conditions. In the Geropotamos River basin, the largest part of the basin including the 'Messara' valley where all the agricultural activity takes place faces humid conditions. Some minor differentiations appear in the mouth of the Geropotamos River near the coastline where sub- humid or dry sub-humid conditions exist during the year. On the other hand, a prolonged part in the northern side of the basin where 'Psiloritis' Mount is located, presents very humid conditions.

4.6 Integrated analysis of Drought

Table 4.18 presents the validation between the reference and the future periods' precipitation data. Table 4.19 depicts the average results of the present and future analysis regarding the drought conditions, while table 4.20 shows the trend analysis results prevailing in each study area. Figures 4.17, 4.18 and 4.19 depict the mapping of SPI values according to the reference periods' data analysis, as well as for the periods 2021-2050 and 2071-2100, under the A1B and B1 scenarios, respectively. The depicting of the future results of the drought analysis was performed by using the mean average values of all the ENSEMBLES RCMs (ensemble mean). The mapping of the reference period was conducted by using the actual observations that were collected from the meteorological stations of the study areas, by following the same downscaling technique. Since in the current areas agriculture is the main economic activity and thus irrigation constitutes a considerable factor and depends greatly on annual reservoirs stage, the 12-month SPI was selected to be used, in order to represent and map the results.

94

Table 4.18. Reference (observed) and future (RCMs simulations) periods' data validation

Ardas Sperchios Geropotamos

A1B B1 A1B B1 A1B B1

EF 0.73 0.82 0.78 0.86 0.85 0.88 std - (reference period) (mm) 71.8 81.6 70.5 std (2001-2100) (mm) 82.7 84.1 93.4 92.0 64.6 74.6 Difference (mm) -10.9 -12.3 -11.8 -10.4 5.9 4.1 Note: std is the Standard Deviation

A first remark as reported by table 4.18 is that the Coefficient of Efficiency shows great effectiveness for the study areas and especially for the Geropotamos and Sperchios River basins. These values were very close to the optimal value of 1.0 and suggest that the RCMs simulated future precipitation values that were used in the current study were almost as efficient as the observed values. Regarding the examination of the standard deviations, the results indicate that the reference and the future periods' data do not present significant variations between each other.

Table 4.19. Drought analysis results occurred by the average means of the examined RCMs

Ardas River basin Sperchios River basin Geropotamos River basin Scenario/Period A1B B1 Scenario/Period A1B B1 Scenario/Period A1B B1 Current 1.14 Current 0.59 Current 0.95 2021-2050 0.52 0.57 2021-2050 -0.12 -0.14 2021-2050 0.26 0.30 2071-2100 0.32 0.47 2071-2100 -0.79 -0.62 2071-2100 0.19 0.20

According to table 4.19, the first fact that can be derived from the 12-month SPI drought analysis is that the drought conditions in all the examined study areas will be more intense in the upcoming years. Namely, in the Ardas River basin, the reference periods' data analysis showed that the area faces 'Moderately Wet' conditions which in the future years will still keep presenting a positive balance. Specifically, during the first examined period of 2021- 2050, the 12-month SPI values will be 0.52 for the A1B and 0.57 for the B1 scenario, respectively. Additionally, for both A1B and B1 scenarios, the Ardas River basin will face the least reductions, in terms of SPI values. Regarding the 2nd examined period of 2071-2100, the 12-month SPI values will keep reducing but the balance will have positive values which are categorized according to the classification scale for the SPI values in table 3.08 as 'Near normal'. B1 scenario shows smaller differentiations from the reference periods' data, while

95

A1B 12-month SPI values especially during the 2nd future period analysis will face harsher reduction. Respecting the Sperchios River basin, the reference periods' data analysis produced a positive 12-month SPI value which can be categorized as 'Near normal'. Nevertheless, the future assessment of drought conditions showed that the Sperchios River basin will face the most intense reduction. Starting from the 1st simulated period, the 12-month SPI values will present a negative balance, but still in 'Near normal' conditions, for both simulated scenarios. Noteworthy is the fact that it is the only case (Sperchios River basin: 2021-2050 SPI drought analysis), where B1 scenario is presented with lower values than A1B scenario. For the 2nd simulated period of 2071-2100, SPI values will continue to decline with the same rate as will happen between the reference periods' data and the 1st simulated period of 2021-2050. In this latter case, B1 scenario reduction will be moderated, compared to A1B scenario which will continue to decrease and it will reach -0.79 in SPI values. This value, although it belongs to the 'Near normal' categorization class, it approaches the limits of the 'Moderately Dry' class. Additionally, for every simulated period and scenario in the Sperchios River basin, the 12-month SPI values will present a negative balance, in terms of drought conditions. Concerning the study area of Geropotamos River basin the 12-month SPI analysis using the reference periods' data produced a value of 0.95 (SPI units). This value belongs to the 'Near normal' class, very close to the limits with 'Moderately Wet' class. Nonetheless, although in both future periods and scenarios the 12-months analysis SPI values are reducing, they still remain in positive values. Regarding the future scenarios of 12-month SPI, a considerable reduction is observed by the 1st examined period, followed by a further - but not so acute- reduction during the 2nd simulation period of 2071-2100.

Table 4.20. Trend analysis results of present and future drought conditions in the study areas

Study area Ardas Sperchios Geropotamos Scenario A1B B1 A1B B1 A1B B1 Z value -0.42 -0.28 -0.54 -0.50 -0.39 -0.33

Mann-Kendall trend analysis for the Ardas River basin for the period 1985-2100 indicated that A1B is following a stronger downward trend than B1 and this is clearly affected by the results of the 12-month SPI for the 2071-2100 period, where the decline of 12-month SPI values for A1B will be more vigorous than B1 by the end of the century. According to the trend analysis that was conducted for the Sperchios River basin regarding the drought conditions, it will face the strongest downward trend amongst the study areas. A1B values will have a negative trend of -0.54, while B1 values will face a negative

96 trend of -0.50. Nevertheless, their negative trend will not be very strong but rather moderate, with the A1B scenario values to be slightly more intense, compared to the B1 values. Mann-Kendall trend analysis for the Geropotamos River basin showed a downward trend. Nevertheless, and in contrast with the Sperchios River basin that faces similar climate conditions in the biggest part of the basin and was presented with the highest downward trends in both applied scenarios, in this certain case the trends analysis results were more temperate than the other study areas. A1B scenario is appeared with the lowest downward trend among the study areas with -0.39, while during the application of the B1 scenario trend analysis, the values were -0.33 which is very close to the values presented for the Ardas River basin that faces continental climate conditions. Subsequently, figures 4.17, 4.18 and 4.19 depict the spatial interpolation of the 12- month SPI drought analysis for the selected study areas. For better understanding, the class 'Near normal' that is mentioned in table 3.08, during the spatial interpolation and the creation of the maps was divided into 2 classes. Values ranging from -0.99 to 0 were given the name 'Dry-Near normal' and values ranging from 0 to 0.99 were categorized as 'Wet-Near normal'.

97

Figure 4.17. Drought analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

98

Figure 4.18. Drought analysis of the Sperchios River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

99

Figure 4.19. Drought analysis of the Geropotamos River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100

100

According to the reference periods' analysis, the SPI values in the Ardas River basin range from 'Dry-Near normal' to 'Extremely Wet'. During the mapping of the 2021-2050 period, almost the entire Ardas River basin will face 'Wet-Near normal' conditions. Regarding the 2nd future period of 2071-2100, the differentiations will be minor, and the spatial distribution of the 12-month SPI values will stay almost in the same levels. Nevertheless, future drought conditions in the Ardas valley will not face drought problems in the upcoming years. During the analysis for the years 2021-2050, the values will be concentrated into 2 categories and the mountainous areas will face 'Wet-Near normal' conditions, while in the lowlands 'Dry-Near normal' conditions will prevail. On the other hand, by the end of the century (period 2071-2100), drought conditions will be intense and the basin will face 'Dry- Near normal' to 'Severely Dry' conditions, for both examined scenarios. Differentiations also appeared during the spatial interpolation of the SPI values between the North and South part of the basin. Drought analysis values range from 'Dry-Near Wet' to 'Severely Wet'. By the 1st period of 2021-2050 the extension of 'Dry-Near normal' conditions will remain the same for both scenarios, while the 'Severely Wet' conditions that prevail in the mountainous areas will considerably decrease. During the 2nd period of 2071-2100, the largest part of the basin will face 'Dry-Near normal' conditions. Some, minor differentiations will appear in the mountainous areas where a prolonged part of the basin will face 'Wet-Near normal' conditions.

101

102

Chapter 5

Discussion

5.1 Selection of the appropriate PET formula

In general, the formulae that can be adopted for the estimation of PET in the Ardas

River basin are PETHam2 and PETHam1, while for the Sperchios and Geropotamos River basin the most appropriate method for the estimation of PET is PETHam2. Furthermore, through the analysis and the results of the current study, air temperature and solar radiation are the most important factors that affect the results of PET. On the other hand, the effect of wind speed in all the study areas was negligible. According to Allen et al. (1998), the drier the atmosphere, the larger the effect on ET. Additionally, for humid conditions, the wind can only replace saturated air with slightly less saturated air and remove heat energy. Consequently, the wind speed affects the evapotranspiration rate to a far lesser extent than under arid conditions where small variations in wind speed may result in larger variations in the evapotranspiration rate. At this point, an important fact that should be mentioned is that since there weren't any measured PET data from the study areas in order to perform the comparison concerning which formula estimates better and is closer to the 'real' data, each PET formula was validated against the average PET for each station examined, and that constitutes a limitation regarding the methodology that was performed in the current study. Several studies as well as software (Kostinakis et al. 2010; Xystrakis and Kostinakis 2011; Xystrakis and Matzarakis 2011; Gebhart et al. 2012) have assessed the sensitivity of the parameters that contribute to the estimation of potential evapotranspiration. The studies highlighted the relative importance of each variable. Saxton (1975) tested the sensitivity of the reference evapotranspiration of the standardized FAO56 Penman-Monteith model during the growing season in northwest China and concluded that shortwave radiation was the most sensitive variable, followed by air temperature. Coleman and DeCoursey (1976) tested the sensitivity of six evaporation and evapotranspiration models and they concluded that the most important parameter at the annual scale is relative humidity (RH), while during summer both temperature and solar radiation are the most important variables. Additionally, wind speed had very small importance at the annual scale.

103

Similar findings were presented by Babajimopoulos et al. (1992), who tested the sensitivity of Penman method and also concluded that temperature and solar radiation are the most important variables during the summer period, whereas the most important parameter in the winter is relative humidity. Eitzinger et al. (2002) analysed the sensitivity of several evapotranspiration methods in different crop-weather models in Austria and concluded that there are many differences between the results, relative sensitivities in the used methods and crucial parameters for their parameterisation. Gong et al. (2006) performed a sensitivity analysis of the Penman-Monteith reference evapotranspiration method in the Changjiang (Yangtze River) basin in China and the results showed that the response of ETref can be precisely predicted under perturbation of relative humidity or shortwave radiation by their sensitivity coefficients. Irmak et al. (2006) performed a sensitivity analysis of the Penman-Monteith method for several regions that are characterized by different climate types, in the U.S. Results indicated that the vapour pressure deficit (VPD) was in general the most sensitive parameter to all locations, although it showed significant variation between the location as well as the period of the year. Additionally, Rs was the dominant force of the equation at humid locations during the summer months. On the other hand, wind speed was very sensitive in semi-arid locations. Estevez et al. (2009) tested the sensitivity of a Penman-Monteith type equation to estimate the reference evapotranspiration in southern Spain. The results showed a large degree of daily and seasonal variability, especially for temperature and relative humidity. Borgmann (2011) tested the sensitivity of 18 different potential evapotranspiration models for six climate stations in Germany and resulted that all investigated PET models were sensitive to significant trends in climate data and therefore they should be validated in a regional context before they are applied to a certain region. Regarding the Greek region, the study performed by Ampas and Baltas (2012) for the wider region of Florina which is located in northern Greece, indicated that the influence of the variables on evapotranspiration is not the same for each period, and also the order that the variables influence evapotranspiration is changing. Finally, Paparrizos et al. (2015) performed a sensitivity analysis of various potential evapotranspiration formulae for Crete Island in South Greece and they concluded that some formulae can produce more accurate results than others. Specifically, the PET equation after Jensen-Haise that is based on mean air temperature and solar radiation and the 1st version of Hamon equation that uses minimum, maximum and mean air temperature as well as Day Length input data proved to estimate with great efficiency PET in an area characterised purely by Mediterranean climate conditions. In relation to these similar studies a common point was that for each type of climate a specific PET empirical method is needed for the estimation of potential evapotranspiration. Furthermore, the results of PET are affected by different parameters amongst various

104 climates as well as seasons; and although the Mediterranean climates are mostly affected by the temperature and radiation based equations, while in temperate climates the 'water' element is introduced and relative humidity (RH) or vapour pressure deficit (VPD) may affect the results, further research is needed before reaching to this conclusion.

5.2 Assessment of precipitation responses

From the integrated analysis it is evident that all the study areas are presented with future decrease in precipitation values. The lowest reduction is observed in the Sperchios River basin, followed by Ardas and Geropotamos River basins, respectively. The similar researches that were quoted in the current study showed a general annual reduction of the precipitation amount that varies from 10-30%, while in the current study results are apparently higher and range between 25-50% for the different areas, scenarios and chronological periods. Many differentiations occurred between the results of the application of A1B and B1 scenarios for Ardas and Sperchios River basins. On the other hand, for the Geropotamos River basin, both scenarios resulted similar values, regarding the future precipitation response. Seasonal analysis showed that in Sperchios and Geropotamos River basins that they are characterized mainly by Mediterranean climate, spring precipitation values will be most affected in every future simulated period and scenario, while in the Ardas River basin which is characterized by a variety of climatic conditions, mainly influenced by the continental conditions, autumn and winter precipitation values will be the most affected. Trend analysis is presented similar for Ardas and Sperchios River basins while the downward trend is more intense for the Geropotamos River basin. Respecting the seasonal analysis, in every case, spring and autumn seasons precipitation are expected to face the greatest reduction, which is very critical for the agricultural field. Especially for the spring precipitation which is the main contributor to the agriculture productivity, a future reduction will create water shortage problems. For this reason, sustainable measures need to be adopted, towards the effective confrontation of water deficiencies. Concerning the changes between the seasonal precipitation over the study areas, the connection between the seasonal variation of future precipitation changes and the seasonal variation of the large scale atmospheric factors being responsible for precipitation needs to be mentioned. In the eastern Mediterranean, the large scale atmospheric mechanisms associated with precipitation formation are not the same between the cold and the warm period of the year. The cold period precipitation is caused mainly by the action of frontal depressions, while in the warm period because of the high static instability associated with

105 the presence of cold upper air masses. This is validated by Feidas et al. (2007), during positive values of the North Atlantic Oscillation Index (NAOI) in winter, Greece becomes significantly cooler and drier as northerly airflow brings cold and dry continental air into the Mediterranean sea. On the other hand, during summer, the centres of action of the NAO shift to the north and the related Azores high extends into central Europe and the Mediterranean basin. Additionally, North Atlantic Oscillation (NAO) plays an important role in the precipitation regime in Greece and the downward trend in precipitation variability is linked principally to a rising trend in the hemispheric circulation models of the NAO. Spatial interpolation of future precipitation is affected by the multivarious background and the topography of the study areas, and shows a great overall reduction. Specifically, for the Ardas River basin, by the end of the century, precipitation levels in the lowlands where the agricultural production takes place will not exceed 400mm. In the Sperchios River basin and specifically in the western part of the basin where the Sperchios River springs are located, precipitation levels are expected to be decreased but these changes will not be so intense compared with the other parts of the basin where precipitation levels will face a much greater reduction. On the other hand, in the south part of the basin and especially around Mount 'Oiti' where the highest point of the current study area is located with an altitude of 2152m, according to the analysis, the precipitation in the worst-case scenario (A1B: 2071- 2100) will be reduced by more than 600mm and it will not exceed 400mm. In every case, precipitation values in the Sperchios valley where the main agricultural production takes place, according to the reference period values analysis are already presented low (0- 400mm), and according to the future scenarios, these values are expected to be more decreased. Nevertheless, according to B1 scenario for the period 2021-2050, a sizable part of the Sperchios valley will still receive up to 800mm in the near future, amount very crucial for the agriculture production. Regarding the spatial interpolation of precipitation for the Geropotamos River basin, the future analysis of precipitation is presented with major differentiations in comparison with the current situation in the highlands, and minor differentiations in the lowlands. Particularly, during the analysis and mapping of precipitation using the existing data, the range of the values varies and up to 600mm fall in the central part of the basin where 'Messara' valley exists while more than 1200mm fall in the mountainous part of the basin where Mount 'Psiloreitis' is located. On the other hand, during the application of the IPCC emission scenarios for the future years, the precipitation will be dramatically decreased. Many recent studies have also contributed to the assessment of present and future precipitation variability over Greece and generally over the Mediterranean region (Repapis 1986; Metaxas et al. 1999; Brunetti et al. 2004; Goubanova and Li 2007; Nastos and Zerefos 2007; 2008; 2009; Nastos et al. 2013a; 2013b; Gocic and Trajkovic 2013; Hertig and

106

Jacobeit 2008; 2014). According to the results of these previous studies, the Mediterranean region has tended towards the decrease of the winter precipitation during the last few decades, mostly starting in the 1970s and proceeding to an accumulation of dry years in the 1980s and 1990s (Schönwiese et al. 1994; Piervitali et al. 1997). Especially over the eastern Mediterranean where Greece is located, decreasing precipitation is also evident in large parts of the sea (Feidas et al. 2007). Schönwiese et al. (1994) reported a pronounced significant trend towards a drier winter climate over the eastern Mediterranean area, for the period of 1961-1990. A general drying is also discernible over most of the south-eastern part of Mediterranean and Greece, which is prominent and statistically significant during the second half of the 20th century (Feidas et al. 2007). As indicated from the results, water planning and management must follow very careful steps. This matter, in combination with the reduced water availability in the certain study areas due to irrigation purposes can result serious water scarcity problems for the local population as well as the agriculture productivity in the future. In order to avoid future drought phenomena due to decreased land precipitation, a general suggestion could be the creation of a dam in Sperchios River basin, similar to the existing ones in Ardas and Geropotamos River basins that will serve irrigation purposes and will also cover the water demand, especially in the summer season where lack of water can create water scarcity problems. Furthermore, the development of an observation-warning system that will control the levels of the water resources (reservoirs, ground- and surface waters) in order to cope with any potential future insufficiencies.

5.3 Assessment of future climate change impacts on the hydrological regime

Initially, after the successful calibration of the model, future meteorological data series were given as inputs and the streamflow was estimated for each area, chronological period and scenario. During the simulation of the reference periods' data, the Ardas River basin was the only study area where the model slightly underestimated the results, compared with the observed data series. In the Sperchios and the Geropotamos River basins, the model output values of the streamflow were slightly higher than the observed values. Nevertheless, the results indicated that in every study area future streamflow will face significant reductions in the upcoming years. Specifically, the Sperchios River basin is expected to lose more than 1/3 of its streamflow by the end of the century, while the Ardas and the Geropotamos River basins will lose almost the 1/5 of their current streamflow. Future decreases in runoff are reflected and confirmed by the decrease in precipitation amount, as it was resulted in the current study (Paparrizos et al. 2016a). Additionally, the amount of water that is stored in the soil profile (SW) will also face strong

107 decreases due to the overall decrease of the water balance amount, but these reductions will be stronger in the Geropotamos River basin. AET plays a very important role in the current study. It is mainly influenced by air temperature and solar radiation (Ampas and Baltas 2012). When hot conditions exist (especially in areas characterized by Csa and Csb climate classification), AET can reach very high levels and this will lead to more amount of uprising moist air, which will be lifted, it will be cooled and the water vapour will condense to form clouds that will re-cater the water bodies through precipitation. Increase of hot conditions and thus of the mean air temperature in the Mediterranean has been reported by the IPCC reports (IPCC 2007), and various studies (Hertig and Jacobeit 2008; Nastos et al. 2013a). According to the outputs of the ARIMA model presented in table 4.11, the irrigation water withdrawal is expected to be raised in the upcoming years, which means that although the streamflow levels in all the examined study areas that are characterized by different climate conditions will be reduced in the future, nevertheless the needs regarding the irrigated water will be increased. On the other hand, the agricultural water withdrawal versus the total water withdrawal will be critically decreased. The increase of the withdrawal water for irrigation is significant, but in the future the ratio regarding the agricultural versus (Vs) the total water withdrawal will follow a downward trend, mainly due to the eventual development of infrastructures, capable of serving, providing and act as a regulator against the insufficient available future amount of water. A comment regarding the implications for the uncertainty associated with ARIMA forecasts in the current study can be the probable insufficiency of the model due to the small sample in proportion to the prediction years could lead to significant deviations from the actual values. Nevertheless, the qualitative characteristics of the model are the same (or nearly the same), which constitutes it as a very satisfactory solution to predict future time series. Summarizing, in order to perform an informative comparison regarding the climate conditions prevailing in Greece and how these will affect the future variations of the hydrological cycle, the three study areas were divided into North (that is represented by Ardas River basin) - Central (Sperchios River basin) - South (Geropotamos River basin) according to their geographical position.

. North: Streamflow will face minor reductions by the mid-21st century that will become major by the end of the century. These areas will be more resistant to the future variations of the hydrological cycle, at least up to a certain period of time. AET will also face significant decrease in areas located in the North. These facts classify the study areas located in northern Greece as the least vulnerable to future change.

108

. Central: Streamflow will face the greatest reductions in the upcoming years. Nevertheless, precipitation is expected to face strong, but compared to the other study areas, the fewer reductions. This fact classifies the study areas located in central Greece as very vulnerable to future change.

. South: Streamflow will face minor reductions in the 1st half of the century that will become more intense by the end of the century. Additionally, precipitation and AET will face the strongest reductions, classifying the certain study area the most vulnerable in terms of future climate changes regarding the water resources.

Numerous researchers have assessed the future changes and responses of the streamflow and the hydrological cycle in similar climates in the Mediterranean region. Indicatively, Arnell (1999) studied the future changes on hydrological regimes in Europe with a continental perspective and concluded that in continental climates, the snowfall will become less important due to higher temperatures, and therefore winter runoff will be slightly increased, while spring runoff will be decreased. Mimikou et al. (2000) performed a simulation in an agricultural catchment in Central Greece and they resulted that seasonal as well as annual runoff will be reduced, while the most significant reduction is expected in mean summer runoff values (May-October). Sanchez-Gomez et al. (2009) studied the future changes in the Mediterranean water budget by an ensemble of RCMs and concluded that already significant changes will start to occur for the period 2030-2050 where the runoff is expected to be decreased by almost 25% that will reach almost 45% by the end of the 21st century. Voudouris et al. (2012) performed an assessment of runoff in future climate conditions in an agricultural area in Crete Island in Greece using the output simulations of the Regional Climate model REMO and establishing a rainfall-runoff model, and resulted drastic mean runoff reductions by 30%. Kalogeropoulos and Chalkias (2013) simulated the future runoff using the Soil and Water Assessment Tool (SWAT) model in Andros Island that faces purely Mediterranean conditions and concluded that the future decrease in rainfall will impact and will cause a significant decrease in mean annual surface runoff. Additionally, a large number of studies have examined potential trends in measures of river discharge during the 21st century, at scales from river basin to global. Some have detected significant trends in some indicators of flow, while others have demonstrated statistically significant links with trends in temperature or precipitation (Milly et al. 2005). Moreover, the majority of these studies attempt to approach this target by scenario-based analyses, which indicate possible trends of future climate evolution based on the assumed (scenarios) trends on forcings (Giorgi et al. 2001; Cayan et al. 2001; Georgakakos 2003; Dettinger et al. 2004). Additionally,

109 few studies have also provided quantitative measures of uncertainty (Koutsoyiannis et al. 2007). In summary, according to previous studies runoff will face strong decreases and downward trends in the upcoming years and decreased land precipitation along with increases in the mean surface temperature will be responsible for these reductions.

5.4 Assessment of present and future Growing Degree Days for agriculture

According to tables 3.09 and the results of the tables 4.15 and 4.16, for the Ardas River basin, not only will the cultivations of asparagus, sugar beet and cherry not face problems, but their cultivation can be expanded in areas with higher altitudes. On the other hand, although the sunflower cultivation requires a high number of GDD until it reaches maturity which according to the reference periods' data analysis can be found only in some, limited areas in the lowlands, in the upcoming years as depicted in figure 4.11, it will find more suitable conditions and its cultivation can be expanded in larger areas. In the upcoming years, the Ardas valley where all the agriculture production takes place will reach up to 2500 growing degree units which will contribute to the conservation of the existing cultivations, or it can even lead to the introduction of new, more pretentious cultivations. The spatial interpolation of the GDD in the Sperchios River basin also indicates that all the existing main cultivations can reach maturity in a shorter period of time even in areas with higher altitudes located in the northern side of the basin. A noteworthy fact here is that although the Sperchios River basin in the springs of Sperchios River on the southern side is affected by mild-wet winters (Csb) according to Köppen-Geiger classification (Peel et al. 2007), nevertheless the GDD of A1B scenario for the months April-October during 2021- 2050 analysis will be similar to the Geropotamos River basin, which is characterized entirely by Mediterranean climate (Csa). Additionally, the Pistachio trees cultivation which constitutes one of the most profitable export products of the current study area can be expanded into higher altitudes and its harvesting period can be shifted from October to September since maturity can be achieved within a shorter period of time. A noteworthy fact during the spatial interpolation of the Geropotamos River basin is that according to figure 4.13 despite the fact that during the reference periods' analysis the areas in the northern side of the basin where high altitudes exist are depicted with high GDD, in the future simulations they are varying between 500-1500 GDD. On the other hand, in these areas which constitute the highest points of the 'Psiloreitis' Mount and are dominated by abrupt slopes, there are very few possibilities for cultivations (at least not extensive cultivations). Lastly, the tomato fields which constitute the most pretentious cultivation in the Geropotamos River basin could be potentially expanded by the period 2071-2100 on areas

110 with higher altitude where the conditions will also be favourable. Nevertheless, harsh topography that exists on the highlands makes this expansion unaffordable for the farmers. Since the current research is amongst the first ones that are focused on the future estimation of GDD under various climate change scenarios and their effect on the local agricultural production, comparisons can be performed only for the reference periods' data. According to a previous research of Matzarakis et al. (2007), Greece is characterized by favourable thermal conditions for the cultivation of high productivity crops. This is validated by the current research which adds the fact that these favourable thermal conditions will increase in the future. Previous similar studies regarding the concept of GDD in the Mediterranean mainly focused on specific cultivations that are of great economic importance for the Mediterranean region. Orlandi et al. (2005; 2014) studied the bioclimatic requirements for the olive tree flowering and the influence of the Olive tree phenology in comparison with the climatic variations, while Köse (2014) studied the phenology and ripening of grape varieties over the Mediterranean. Canavar and Ali Kaynak (2010) observed the sunshine radiation effects on peanut pod yield and growth in combination with the GDD. Matzarakis et al. (2007) made a first attempt to interpolate the GDD over Greece and link them with the main cultivations that exist in Greece using data for a 10-year period from 1978-1987. According to the results, the GDD for the mean growing season in Greece ranges from about 1600 GDD units in the northern and mountainous areas of the country to more than 2900 GDD units in the lowlands. These limits exceeded the crop maturity requirements of most crops that exist in the Greek territory. Koufos et al. (2014) studied the temporal evolution of Growing Degree Days in the main winegrape regions of Greece and the results indicated an overall statistically significant positive trend of all indices. Finally, Paparrizos et al. (2016d) used the Growing Degree Days and precipitation as assessment factors for future responses in agriculture in Evros region, in Greece. Specifically for the GDD units, these are expected to be increased in the upcoming years and the cultivation period may be shifted from April- October to April-September. On the other hand, harsh topography will act as an inhibitor towards the expansion of the existing cultivations onto higher altitudes. Through the spatial distribution of future GDD in the study areas it is evident that all the study areas are presented with future increase in GDD values. The biggest increase is expected in the Sperchios River basin, followed by the Ardas and the Geropotamos River basins, respectively. Although there were some minor differentiations between the two applied scenarios, their results were almost identical in most of the cases, which is another fact that validates the results of the current study. Another noteworthy fact is that during the sub-period analysis, the Sperchios and the Geropotamos River basins that are characterised mainly by Mediterranean climate, in both future periods of 2021-2050 and 2071-2100 under

111 both applied scenarios presented critical differentiations between April-October and April- November results. In summary, the main cultivation period in all the study areas will be reduced and shifted from April-October to April-September, especially in 1-year cultivations. This fact will directly affect (and reduce) the production costs, which in combination with the new technologies that occur every day will further decrease the expenses. Furthermore, a part of the surplus can be invested towards the improvement of the quality as well as the quantity of the existing crops in order to raise their productivity, under the frame of sustainable development. Additionally, the reduction of the cultivation period can lead to the mitigation of the amount of energy required to provide products and services and contribute towards the efficient energy use. Moreover, new, more demanding but also more profitable and efficient cultivations can be introduced in the certain study areas as well as areas characterised by similar climate conditions in the upcoming years. At this latter case though, regardless the fact that as it was indicated by the results of the current study that the cultivation range is expected to be expanded in the upcoming years in all the study areas, the multivarious background and the complex terrain of the areas act as an inhibitor towards the actualization of the specific scheme. In other words, although an expansion is possible to be achieved, nevertheless it would have affected not directly the cultivations but instead the farmers as they would have needed to perform this expansion in areas with rough topography and so the cultivation costs would have dramatically increased. Summarizing, all the similar studies indicated that GDD units are, or will follow a positive trend. On the other hand, topography will act as an inhibitor towards the expansion, as it was resulted in similar studies regarding the GDD (Paparrizos et al. 2016d).

5.5 Assessment of aridity conditions

Regarding the integrated analysis of the Aridity Index in the examined study areas, special attention needs to be given firstly to the Geropotamos River basin. The under examination study area is characterized by the presence of Mount 'Psiloritis' (2.456m), which is the highest mountain of Crete Island. Mount 'Psiloritis' is dominated by abrupt slopes, a phenomenon that exists in the whole territory of Crete Island. Additionally, the Geropotamos River basin is generally a basin with 'small' distances from the watershed limits to the coast line (which is the mouth of the water basin), especially in the northwest part where Mount 'Psiloritis' lies. Seasonal analysis in the Geropotamos River basin is also affected by the topography of the area as in the North part where higher altitudes exist, the AI is presented with more temperate conditions, compared with the South part of the basin where the

112 conditions are more severe and semi-arid or arid conditions appear for the spring and summer months following the Mediterranean climate conditions. Nonetheless, trend analysis showed that the part of the northern basin with the high altitudes tends to increase its aridity conditions. The map that depicts the fluctuation of AI for the spring season (figure 4.14) is the most characteristic in order to interpret the aridity conditions prevailing in Ardas River basin. According to the spatial interpolation of the AI for the spring season, four classes of AI classification exist with conditions ranging from humid (in the upper mountainous part of the basin) to semi-arid in the mouth of the basin. Regarding the trend analysis of aridity for the Ardas River basin, the lower parts present a very high downward trend while the mountainous west part of the basin where the sources of the Ardas River are located shows a mighty upward trend. This specificity tends to moderate the aridity conditions prevailing in the Ardas River basin within the next years. Concerning the Sperchios River basin, as shown in figure 4.15 in some parts of the water basin, arid-conditions appeared. Sperchios River basin has a Mediterranean climate (Csa). Areas with this type of climate receive almost all of their precipitation during the winter, autumn and spring seasons and may go anywhere from 4 to 6 months during the summer without having any significant precipitation (Peel et al. 2007). This is verified by the fact that spring and summer seasons face dry sub-humid and semi-arid conditions respectively, while autumn is the season that faces the most humid conditions of the year. The low values of the AI especially on the certain seasons which constitute the main and the most critical face of the cultivation period point out the sententiousness of the issue and special attention needs to be given in order to avoid water scarcity problems that will have great impacts on the local population related with the agriculture activities. Differentiations during the spatial interpolation of the AI also appeared between the North and South part of the basin. The South part of the basin is characterized by high altitudes, while in the North part of the basin, moderate altitudes exist. This specificity affects the analysis as in figure 4.15 the results in the southern part are more temperate compared with the northern part of the basin. Generally, all the study areas showed an upward trend regarding the aridity conditions in the mountainous areas while the trend completely changes and follows a downward tendency in the lowland areas where all the agriculture activity takes place (plain areas). It is evident that in this case also, elevation, as well as the geographic position play a very important role in the creation of the climatic conditions prevailing in the results of the integrated analysis of aridity in the study areas. Furthermore, in locations with very abrupt slopes, the trend analysis indicated that the trend follows an upward intense tendency in all the study areas. A noteworthy fact is that during the trend analysis, all the study areas were presented with heterogenic trend although their size is characterized as medium.

113

Additionally, another noticeable fact is the existence of large dams in Ardas and Geropotamos River basins that serve irrigation purposes. These dams can eliminate the difference between water supply and demand, especially in the summer months when the need for available water for irrigation is increased and the conditions are semi-arid or arid, as it was resulted from the current research. The water demand in the lower areas is higher than in other parts of the basin and it is apparent that if there is no rational management of the available water resources, drought conditions will occur and have great impacts on agriculture productivity. Similar studies have been conducted in Greece and generally in the Mediterranean region towards the assessment of aridity conditions and several indicators and techniques have been proposed for the estimation and quantification of aridity. All these studies have used several statistical techniques to analyse spatio-temporal parameters of drought episodes and assess the conditions regarding the aridity (Nastos and Zerefos 2009). Nastos et al. (2013a) performed an application of the Aridity Index on a large scale for the whole Greek territory and Türkes (2003) for Turkey. The results of the above mentioned studies compared with the results from the current research, are similar, but it must be mentioned that these studies are different comparing to the current research. Specifically, the accurate estimation and mapping of aridity must take into account several factors. It is difficult to estimate the aridity conditions in a sizeable area such as Greece, which presents large and complex elevation variation, if less data is existing as inputs from the current meteorological network. Additionally, in relation to all the above mentioned indices and studies, the technique for the estimation of PET that was followed in the current study that takes into account multiple variables as inputs (RH, Tmean, Tmax, Tmin, Tdew, PREC, Rs, Rn, Ra, and WS) constitutes an innovation and validates that results of the current study compared with other similar studies and also highlights the fact that all future studies should be performed at least on a mesoscale, using sufficient inputs parameters, in order to produce reliable results. PET plays a key role in the current study as it affects the fraction of the Aridity Index and results low values in many cases especially during the spring and summer periods. During the winter season the exact opposite occurs. In the Ardas River basin where the climate is continental and the higher amounts of rainfall occur during the summer months, although PET has high values, the numerator of the fraction has also high values so aridity conditions are reduced. Same conditions occur during the winter season. At this latter case, however, there aren't very humid conditions as the levels of the winter precipitation are not so high. Therefore the results of the AI are strongly influenced by the behaviour of PET. The results are confirmed as Crete Island during the data analysis appeared in many cases with almost zero rainfall from May to October. The use of a different index and method than this of

114 the current research it would have gave extremely arid conditions in the Geropotamos River basin without taking into account various topographical and meteorological variables whose combination can serve as a regulator and prevent possible aridity and drought conditions.

5.6 Assessment of present and future drought conditions

Generally, all study areas will face relatively severe drought conditions in the upcoming years. The conditions will be more intense in the lowland areas (plain areas) where all the agricultural activity takes place. As depicted by the figures, it is evident that elevation and broadly the topography plays a very important role in the creation of different climatic conditions and directly affects the spatial analysis of the drought conditions. In any case, the 12-month SPI drought analysis indicated that the SPI values will follow a downward trend. As stated in table 4.19, the Sperchios River basin will face the harsher reductions in SPI values and it constitutes the only study area where the drought balance (in SPI values) will be negative in the upcoming years. Additionally, according to table 4.20, trend analysis indicated that the Sperchios River basin will face the strongest downward trends in both applied scenarios, followed by the Ardas and lastly the Geropotamos River basins. Especially for the Geropotamos River basin, although it is characterized by Mediterranean climate and someone could have expected stronger downward trends, nevertheless the detected trends were clearly influenced by the complex topography that it is presented in the Geropotamos River basin where the distances between the mountainous areas and the lowlands are very small and thus this acts as an inhibitor towards decreased land precipitation, reduced runoff and extent drought conditions. Hence, special attention needs to be given in the sustainable management of rainfall and snowfall amount prevailing in the Sperchios River basin. For this reason, the creation of a water storage dam in the mountainous or the semi-mountainous area on the watershed is mandatory. Additionally, this dam can contribute towards the reduction of Potential Evapotranspiration which is very crucial for agriculture, since the water will not be evaporated or transpirated in the lowlands, as it is currently happening. Instead, it can act as a regulator and face possible drought conditions that are very likely to appear as it is inferred from the analysis of the current study. Many studies have contributed towards the assessment of drought conditions over Greece and generally the Mediterranean region and several indicators and techniques have been proposed for the estimation and quantification of drought conditions. However, in recent years, shortcomings in existing indicators are highlighted and this becomes a quest for the emergence of new indicators and studies (Tsakiris and Vangelis 2005), as well as software (Tigkas et al. 2015). Studies implementing various drought indexes such as Palmer Drought

115

Severity Index (PSDI) and Reconnaissance Drought Index (RDI) (Vangelis et al. 2013), the Standardized Precipitation Index (SPI) (Tsakiris and Vangelis 2004; Loukas et al. 2008; Karavitis et al. 2011; Vroxidou et al. 2013) have been conducted. All these studies have used several statistical techniques to analyze spatio-temporal parameters of drought episodes in Greece (Nastos and Zerefos 2007; 2008; 2009). Mediterranean climate presents differentiations between eastern and western regions. Drought conditions are mainly established at the southern and eastern regions of the Mediterranean basin (Nastos et al. 2013b). In Italy, a drought research indicated that in the last 30 years the drought periods have become more severe and extreme due to decreased land precipitation, mainly in the south part of the country (Piccarreta et al. 2004). In Turkey, drought has been a recurrent phenomenon for the last decades. A warning trend began in the early 1990s and has continued until recent years. The drought conditions prevailed in over the last 2-3 decades in Turkey are related to changes in the weather patterns in the Atlantic region, specifically in the variations in the North Atlantic Oscillation (Komuscu 2001). In Cyprus, droughts are very frequent events with adverse impacts on many fields such as the economy, the environment and the agricultural productivity. According to Pashiardis and Michaelides (2008) and Michaelides and Pashiardis (2009), during the period 1971-2008 Cyprus suffered from nine drought events resulting in the emptiness of the dams. In comparison with the present study's results, similar studies that were performed in various regions of Greece have indicated that the drought conditions in Greece are expected to be more severe in the following years. A study performed by Vasiliades et al. (2009) for Lake Karla in Thessaly, central Greece, near Sperchios River basin using the SRES A2 and B2 scenarios for the future years of 2021-2050 & 2071-2100 reported that drought conditions will be significantly increased in drought severity. Additionally, a drought analysis that was performed in Crete Island by Vroxidou et al. (2013) for the years 1973-2099 concluded that during the period 2035-2099 the drought phenomena will become more intense, as mean precipitation is projected to decrease, whereas wet conditions are completely absent. These results point out the sententiousness of the issue and special attention needs to be given in order to avoid water scarcity problems that will have great impact on the local population related with agricultural activities. Moreover, water demand in lowlands is higher than in other parts of the basin, and it is apparent that if there is no rational management of the available water resources, drought impacts will be disastrous. The emergence of drought conditions is possible, thus the water resources planning and management must follow very careful steps.

116

5.7 Synthesis report

In an attempt to combine the findings and as the focusing point of the current study was the hydrological cycle, the various aspects that define it were assessed. The -never ending- water circulation from the land to the clouds, and back to the land again was addressed under a complex and comprehensive approach. The study was characterised by the existence of limited data derived from the network of climatological stations in the study areas, that constituted the spatial distribution of the examined climatological variables very challenging. The proposed downscaling technique can be a helping hand when operating in the mesoscale for various types of climate. The spatial distribution of the examined variables in the current study (i.e. precipitation, aridity and drought conditions, GDD) is in every case dominated by the harsh topography of the study areas. Although the GDD analysis indicated that topography will act as an inhibitor towards the expansion of the cultivations onto higher altitudes, nevertheless the spatial distribution also showed that precipitation and drought conditions will not be so severe onto higher altitudes. On the other hand, although in the lowlands the cultivations will find more favourable conditions in the future, nevertheless, precipitation will be significantly decreased and drought conditions will be more severe. The water resources will be significantly influenced by that fact as they expected to be critically reduced. In general, the climatic conditions are expected to be more severe in areas with lower latitude, while they will become less severe as the latitude is increased. Figure 5.01 depicts a synthesis report graph that includes a summary of the results that were excluded from the current study.

117

Figure 5.01. Synthesis report graph of the results from the current study

118

Chapter 6

Conclusions and Outlook

6.1 Conclusions

The current study was focused on the present as well as the future assessment of the aspects of the hydrological cycle for three indicative areas widespread in Greece. Initially, a sensitivity analysis and inter-comparison of PET formulae was performed and the most effective PET empirical method as well as the meteorological parameters that influence the estimation of PET in each study area were defined. Afterwards, a present and future assessment of precipitation responses, aridity and drought conditions, and runoff was conducted in order to depict the future variations and the changes of the aspects of the hydrological cycle, and lastly the Growing Degree Days concept was implement in order to link the changes in the aspects of the hydrological cycle with the agriculture productivity. Future simulations from the ENSEMBLE project were obtained, then analysed regarding their statistical significance and applied for the future periods 2021-2050 and 2071- 2100, under the A1B and B1 scenario, accordingly. Topographical data as well as hydrological data for the calibration of the ArcSWAT model were also obtained and given as inputs in order to simulate and assess the future response of runoff. Various maps were created that depict the present and future responses of the aspects of the hydrological cycle in areas that face different climate conditions, which visualize and give fundamental ideas about the climate variations in the selected study areas. Moreover, they constitute a helpful tool for all the stakeholders and they can be notified free of charge, while their understanding is possible, even for non-expert personnel. Geostatistical interpolation has become an important tool in applied climatology because it is based on the spatial variability of the variables of interest and makes it possible to quantify the estimation uncertainty (Maris et al. 2013). The technique of the spatial interpolation that was implemented and applied in the current study where a combination of the Ordinary Kriging method and the multi-linear regression analysis were coupled, allowed several parameters to be taken into account. The initial default application of the Ordinary Kriging method through ArcGIS does not take into account any topographic, climatological

119 and generally any additional parameters that affect the examined (or under interpolation) variable. Especially in areas with limited data like the current study areas, interpolating climatological variables from point stations is insufficient and most of the times misleading. Through the creation of a grid of sample points that cover the entire study area from the lowest to the highest point and by comprehending the multi-linear regression techniques, all the factors that affected each time the spatial interpolation of the examined variables in the current study (precipitation, SPI, AI, GDD, etc.) were included, and high-resolution maps were created. During the spatial interpolation and downscaling technique that was performed in the current study, statistical significance was preserved by implementing the p-value. P-value constitutes a great advantage during statistical analyses particularly for the complex terrain of Greece, in order to avoid significantly misleading conclusions by using insignificant parameters. The effect of topography in the spatial interpolation plays the most significant role and affects the spatial distribution of the examined variables in the current study. The study areas are characterized by complex topography, abrupt slopes and high elevation variability. Rough topography and the complex climate conditions that prevail in the study areas in combination with diverse social, economic and environmental aspects render the Mediterranean region as one of the most vulnerable, in terms of climate change. Additionally, a fact that needs to be mentioned is that although there has been some improvement in the latest years in the simulation of continental-scale patterns of precipitation since the 4th assessment report (AR4) that was published in 2007 by the Intergovernmental Panel on Climate Change (IPCC), at regional scales, precipitation still cannot be simulated precisely, and the assessment is hampered by observational uncertainties. Complex topography and observational uncertainties are the main reasons according to which all the upcoming studies need to focus on a mesoscale and local scale as it was performed in the current study. The purpose of focusing and working on a regional scale constitutes a great advantage because it gives the opportunity to the researcher to implement more detailed components such as topography of the study area, meteorological observations from a denser meteorological network, etc.. On the other hand, working on a larger scale requires a large amount of input data in order to produce satisfactory results as well as a combination of techniques, modes, and various procedures that can often lead to overestimation or, even worse, to an underestimation of the events. Having in mind all these facts, in combination with all the recent findings in the literature that have proved that the effects of the climate change are significant, the researchers are called to undertake crucial decisions regarding the sustainable development of the environment.

120

Runoff is expected to be greatly reduced in areas where complex climate conditions exist and are mainly affected by the Mediterranean climate, while areas characterized by continental climate will be more resistant to future changes. Furthermore, decreased land precipitation is one of the main reasons that contributes to that fact. On the other hand, due to future increases in mean air temperature, AET values will not face strong variations and will partly set against this scheme. Increased mean air temperature will lead to the expansion of the GDD units as well as their extension as the local cultivations will find favourable conditions in larger areas and increase their productivity. This can potentially lead to changes in land use/land cover, but their extensive expansion will governed by the complex topography and will face difficulties. On the other hand, in the existing cultivations maturity can be achieved in a shorter period of time and this will further reduced the cultivations costs. Additionally, the decrease in the cultivation period will have a direct effect on the efficient energy use as well as the protection, preservation and sustainable development of the environment, since irrigation dominates and constitutes the main cause of water scarcity problems in the certain study areas. Concluding, since the climate in Mediterranean is one of the most vulnerable in terms of climate change, great caution is needed when making conclusions. as it is highlighted by the results, the climate changes will be severe in the upcoming years in all the study areas, regardless the climate conditions prevailing in each area. For this reason, water resources management must take serious actions in order to avoid future water problems. The Directive 2000/60/EC of the European Parliament (European Parliament and Council 2000) clearly states the framework of how all the interesting parties should act. Therefore, a strong an efficient cooperation is needed towards the water resources management. Moreover, although the Sperchios River basin will be expected to face the greatest reductions regarding the streamflow, nevertheless it is the only area without a large reservoir that can act as a regulator, especially in periods when the streamflow is critically reduced (Dry period). Thus, a creation of a dam that will serve irrigation purposes is mandatory. Regarding the rest study areas, it is critical to examine if the existing water volume of the reservoirs is sufficient to meet the needs of the agricultural production.

6.2 Outlook

Further research can be focused on the present and future seasonal assessment and analysis of the conditions prevailing in between the PET values, precipitation, drought and aridity conditions, and runoff, especially during the dry period when the greatest part of the agriculture productivity takes place in the current study areas. Hence, more comprehensive explanations regarding the seasonal variation of the aspects of the hydrological cycle can be

121 obtained. Regarding the GDD, a critical point that needs to be further analyzed is the GDD and their variation in combination with future precipitation changes. Although the current study indicated that the conditions will be more favourable, nevertheless, the evident decrease in land precipitation might act as an inhibitor towards the future and sustainable growth and development of the existing cultivations in the study areas. Additionally, a future research can focus on each species individually and perform an analysis in relation with the micro-climate that exists and affects each cultivation. A system can be manufactured that will receive air temperature data as inputs directly at the same point on real-time conditions, and it will estimate the GDD units to maturity, giving the opportunity to the agriculture section to optimize its productivity. Finally, an adequate meteorological network needs to be created; otherwise all research towards this direction be governed by these deficiencies. In any case, the findings of the current study must be taken into account as a guide for future climate conditions. They can be a helping hand for all the farmers and various stakeholders, as they can obtain significant and accurate information so they can further adjust their adaptation systems and manage their agriculture procedures towards the rational management of the agricultural water resources. In order to avoid drought and aridity conditions, flooding phenomena or water scarcity problems, the science of water resources management in combination with meteorology and climatology are called to play a major role in the era of climate change and should design and implement numerous technical projects, infrastructures, and perform further research which today are more than ever necessary in order to prevail against the effects of climate change.

122

References

Ackermann, W. C., Colman, E. A., Ogrosky, H. O., 1955: From ocean to sky to land to ocean. In: U.S. Department of Agriculture Yearbook 1955. U.S. Department of Agriculture, Washington D.C.

Agnew, C., Anderson, W., 1992: Water Resources in the arid realm. Routledge, London, U.K.

Alcamo, J., Flörke, M., Marker, M., 2007: Future long-term changes in global water resources driven by socio-economic and climatic change. Hydrological Sciences Journal, 52, 247–275.

Alcamo, J., Döll, P., Henrichs, T., Kaspar, F., Lehner, B., Rösch, T., Siebert, S., 2003: Global estimates of water withdrawals and availability under current and future “business-as-usual” conditions. Hydrological Sciences Journal, 48, 339–348.

Alexandris, S., Kerkides, P., Liakatas, A., 2006: Daily reference evapotranspiration estimates by the 'Copais' approach. Agricultural Water Management, 82, 371-386.

Allan, C., Ferguson, L., 2014: Physiology of Fruit Growth as a Function of Heat in Commercial Pistacia vera Species. In: Proceedings of American Society for Horticultural Science, July 28-31, 2014, Orlando FL.

Allen, R., Pereira, L., Raes, D., Smith, M., 1998: Crop evapotranspiration - Guidelines for computing crop water requirements. Journal of Irrigation and Drainage Engineering, Paper 56, Rome: Food and Agriculture Organization (FAO), 300 p..

Ampas, V., 2010: Research and estimation of meteorological parameters with direct impact on agriculture. Ph.D. Thesis, Aristotle University of Thessaloniki, Greece.

Ampas, V., Baltas E., 2012: Sensitivity analysis of different evapotranspiration methods using a new sensitivity coefficient. Global NEST Journal, 14 (3), 335-343.

Anderton, S., Latron, J., Gallart, F., 2002: Sensitivity analysis and multi-response, multi- criteria evaluation of a physically based distributed model. Hydrological Processes, 16, 333-353.

Arnell, N. W., 1999: The effect of climate change on hydrological regimes in Europe: a continental perspective. Global Environmental Change, 9, 5-23.

Arnell, N. W., 2004: Climate change and global water resources: SRES emissions and socio economic scenarios. Global Environmental Change, 14, 31–52.

123

Arnold, J. G., Srinivasan, R., Muttiah, R. S., Williams, J. R., 1998: Large area hydrologic modeling and assessment Part I: model development. Journal of the American Water Resources Association, 34(1), 73-89.

ASCE Task Committee on Standardization of Reference Evapotranspiration of the Environmental and Water Resources Institute, 2005: The ASCE standardized reference evapotranspiration equation. In: Allen, R., Walter, I. A., Elliot, R. L., Howell, T. A., Itenfisu, D., Jensen, M. E., Snyder, R. L., (eds.), ASCE, Reston, VA, 172pp.

Babajimopoulos, C., Antonopoulos, Β., Grigoriadis, D., Ilias, Α., 1992: Sensitivity analysis of the Penman method. In: Proceedings of 5th conference of H.Y.U., p. 132-140.

Bacchi, B., Kottegoda, N., 1995: Identification and calibration of spatial correlation patters of rainfall. Journal of Hydrology, 165, 311-348.

Baker, J. T., Pinter, P. J. Jr., Reginato, R. J., Kanemasu, E. T., 1986: Effects of temperature on leaf appearance in spring and winter wheat cultivars. Agronomy Journal, 78, 605- 613.

Barrios, J. E., Rodríguez-Pineda, J. A., Benignos, M. D., 2009: Integrated river basin management in the Conchos River basin, Mexico: A case study of freshwater climate change adaptation. Climate and Development, 1(3), 249-260.

Bates, B. C., Kundzewicz, Z. W., Wu, S., Palutikof, J. P., 2008: Climate Change and Water. Technical Paper of the Intergovernmental Panel on Climate Change. Geneva: IPCC Secretariat, 210 p.

Bauer, A., Garcia, R., Kanemasu, E. T., Blad, B. L., Hatfield, J. L., Major, D. J., Reginato, R. J., Hubbard, K. G., 1988: Effect of latitude on phenology of Colt winter wheat. Agricultural & Forest Meteorology, 44, 131-140.

Beek, E., Stein, A., Jansen, L., 1992: Spatial variability and interpolation of daily precipitation amount. Stochastic Hydrology and Hydraulics, 6, 304-320.

Beres, D. L., Hawkins, D. M., 2001: Plackett-Burman technique for sensitivity analysis of many-parametered models. Ecological Modelling, 141, 171-183.

Beven, K., 1979: A sensitivity analysis of the Penman-Monteith actual evapotranspiration estimates. Journal of Hydrology, 44, 169-190.

Borgmann, H., 2011: Sensitivity analysis of 18 different potential evapotranspiration models to observed climatic change at German climate stations. Climate Change, 104, 729- 753.

124

Botai, C. M., Botai, J. O., Muchuru, S., Ngwana, I., 2015: Hydrometeorological Research in South Africa: A review. Water, 7, 1580-1594.

Box, P., Jenkins, G. M., 1976: Time series analysis: Forecasting and control. San Francisco, CA: Holden-day Inc.

Braun, O., Lohmann, M., Maksimovic, O., Meyer, M., Merkovic, A., Messerschmidt, E., Reidel, E., Turner, M., 1999: Potential impact of climate change effects on preferences for tourism destinations: a psychological pilot study. Climate Research, 11, 2477– 2504.

Brunetti, M., Maugeri, M., Monti, F., Nanni, T., 2004: Changes in daily precipitation frequency and distribution in Italy over the last 120 years. Journal of Geophysical Research, 109, D05102.

Calder, I.R., 1990: Evaporation in the Uplands. John Wiley and Sons, Chichester, United Kingdom.

Canavar, Ö., Ali Kaynak, M., 2010: Growing degree day and sunshine radiation effects on peanut pod yield and growth. African Journal of Biotechnology, 9(15), 2234-2241.

Caprio, J., 1974: The solar thermal unit concept in problems related to plant development ant potential evapotranspiration. Phenology and seasonality modeling. In: Lieth, H., (eds.), Springer, New York, 353–364.

Cayan, D. R., Kammerdiener, S. A., Dettinger, M. D., Caprio, J. M., Peterson, D. H. 2001: Changes in the onset of spring in the western United States. Bulletin of American Meteorological Society, 82, 399-415.

Chang, C., Lo, S., Yu, S., 2005: Applying fuzzy theory and genetic algorithm to interpolate precipitation. Journal of Hydrology, 314, 92-104.

Chen, C. C., Gillig, D., McCarl, B. A., 2001: Effects of climatic change on a water dependent regional economy: a study of the Texas Edwards aquifer. Climate Change, 49(4), 397– 409.

Cho, S. M., Lee M. W., 2001: Sensitivity considerations when modelling hydrologic processes with digital elevation model. Journal of the American Water Resources Association, 37(4), 931–934.

Christensen, J. H., Christensen, O. B., Lopez, P., van Meijgaard, E., Botzet, M., 1996: The HIRHAM4: Regional atmospheric climate model. Danish Meteorological Institute, Scientific Report 96-4.

125

Chua, S., Bras, H., 1982: Optimal estimators of mean areal precipitations in regions of orographic influence. Journal of Hydrology, 57(1-2), 23-48.

Coleman, G., DeCoursey, D. G., 1976: Sensitivity and model variance analysis applied to some evaporation and evapotranspiration models. Water Resources Research, 12(5), 873-879.

Cross, H. Z., Zuber, M. S., 1972: Prediction of flowering dates in maize based on different methods of estimating thermal units. Agronomy Journal, 64, 351-355.

Cutforth, H. W., Shaykewich, C. F., 1989: Relationship of development rates of corn from planting to silking to air and soil temperature and to accumulated thermal units in a prairie environment. Canadian Journal of Plant Science, 69, 121-132.

Dai, A., Trenberth, K. E., 2002: Estimates of freshwater discharge from continents: Latitudinal and seasonal variations. Journal of Hydrometeorology, 3, 660-687.

Dai, A., Trenberth, K. E., Qian, T., 2004: A global data set of Palmer Drought Severity Index for 1870-2002: A relationship with soil moisture and effects of surface warming. Journal of Hydrometeorology, 5, 1117-113.

Dalezios, N., R., Bartzokas, A., 1995: Daily precipitation variability in semiarid agricultural regions. Hydrological Sciences Journal, 40(5), 569-585.

Dalezios, R. N., Loukas, A., Vasiliades, L., Liakopoulos, E., 2000: Severity-Duration- Frequency analysis of droughts and wet periods in Greece. Hydrological Sciences Journal 45(5), 751-769.

Davidson, H. R., Campbell, C. A., 1983: The effect of temperature, moisture and nitrogen on the rate of development of spring wheat as measured by degree days. Canadian Journal of Plant Science, 63, 833-846.

Dettinger, M. D., Cayan, D. R., Meyer, M. K., Jeton, A. E., 2004: Simulated hydrologic responses to climate variations and change in the Merced, Carson, and American river basins, Sierra Nevada, California, 1900-2009. Climatic Change, 62, 283-317.

Dingman, S. L., 1994: Physical Hydrology. Macmillan Publishing Co., New York, NY.

Dingman, S. L., 2002: Physical Hydrology. 2nd ed. Prentice Hall, Upper Saddle River, NJ.

Edwards, D. C., McKee, T. B., 1997: Characteristics of the 20th century drought in the United States at multiple time scales. Climatology Report No. 97-2, Department of Atmospheric Science, Colorado State University, Fort Collins, CO

Edwardson, S. E., Watt, D. L., 1987: GROWTH STAGE: Using growing degree days to predict the Haun scale of spring wheat. Applied Agricultural Research, 2, 224-229.

126

Eitzinger, J., Marinkovic, D., Hösch, J., 2002: Sensitivity of different evapotranspiration calculation methods in different crop-weather models. In: Proceedings of iEMSs, June 24-27, 2002, Lugano, Switzerland.

Estevez, J., Gavilan, P., Berengena, J., 2009: Sensitivity analysis of a Penman-Monteith type equation to estimate reference evapotranspiration in southern Spain. Hydrological Processes, 23, 3342-3353.

European Parliament and Council, 2000: Directive 2000/60/EC of the European Parliament and of the council of 23 October 2000. Official Journal of the European Communities, L327, 1-72.

FAO, 1993: Forest Resources Assessment 1990. Tropical Countries. Forestry Paper 112. Rome: Food and Agriculture Organization of the United Nations (FAO).

FAO, 2015: AQUASTAT Main Database. Food and Agriculture Organization of the United Nations (FAO). [accessed: 16/05/2016]

FAO, 2016: AQUASTAT website. Food and Agriculture Organization of the United Nations (FAO). [accessed: 16/05/2016]

Feidas, H., Karagiannidis, A., Keppas, S., Vaitis, M., Kontos, T., Zanis, P., Melas, D., Anadranistakis, E., 2014: Modeling and mapping temperature and precipitation climate data in Greece using topographical and geographical parameters. Theoretical & Applied Climatology, 118, 133-146.

Frisch, P., Alexander, L. V., Della-Marta, P., Gleason, B., Haylock, M., Klein Tank, A. M. G., Peterson, T., 2002: Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate Research, 19, 193-212.

Gallagher, J. N., 1979: Field studies of cereal leaf growth: I. Initiation and expansion in relation to temperature and ontogeny. Journal of Experimental Botany, 30, 625-636.

Gambolati, G., Volpi, G., 1979: A conceptual deterministic analysis of the kriging technique in hydrology. Water Resources Research, 15(3), 625-629.

GCOS, 2003: The Second Report on the Adequacy of the Global Observing Systems for Climate in Support of the UNFCCC. GCOS-82, WMO/TD No. 1143, Global Climate Observing System, 74 pp.

GCOS, 2004: GCOS Implementation Plan for the Global Observing System for Climate in support of UNFCCC. GCOS-92, WMO/TD 1219, Global Climate Observing System, 136 pp.

127

Gebhart, S., Radoglou, K., Chalivopoulos, G., Matzarakis, A., 2012: Evaluation of potential evapotranspiration in central Macedonia by EmPEst. In: Helmis, G. G., Nastos, P. T., (eds.) Advances in Meteorology, Climatology and Atmospheric Physics. Springer Atmospheric Sciences, Springer, 1, 451-456.

Gemmer, M., Becker, S., Jiang, T., 2004: Observed monthly precipitation trends in China 1951-2002. Theoretical and Applied Climatology, 77, 39-45.

Georgakakos, K. P., 2003: Probabilistic climate-model diagnostics for hydrologic and water resources impact studies. Journal of Hydrometeorology, 4, 92-105.

Gibelin, A. L., Deque, M., 2003: Anthropogenic climate change over the Mediterranean region simulated by a global variable resolution model. Climate Dynamics, 20(4), 327- 339.

Gilmore, E. C., Rogers, J. S., 1958: Heat units as a method of measuring maturity in corn. Agronomy Journal, 50, 611-615.

Giorgi, F., Lionello, P., 2008: Climate change projections for the Mediterranean region. Global and Planetary Change, 63(2–3), 90–104.

Giorgi, F., Whetton, P. H., Jones, R. G., 2001: Emerging patterns of simulated regional climatic changes for the 21st century due to anthropogenic forcings. Geophysical Research Letters, 28, 3317-3321.

Gocic, M., Trajkovic, S., 2013: Analysis of precipitation and drought data in Serbia over the period 1980-2010. Journal of Hydrology, 494, 32-42.

Goodess, C., Palutikof, J., 1998: Development of daily rainfall scenarios for southeast Spain using a circulation-type approach downscaling. International Journal of Climatology, 10, 1051–1083.

Gong, L., Chong-yu, X., Chen, D., Halldin, S., Chen, Y. D., 2006: Sensitivity of the Penman- Monteith reference evapotranspiration to key climatic variables in the Changjiang (Yangtze River) basin. Journal of Hydrology, 329, 620-629.

Goubanova, K., Li, L., 2007: Extremes in temperature and precipitation around the Mediterranean basin in an ensemble of future climate scenario simulations. Global & Planetary Change, 57, 27–42.

Goyne, P. J., Woodruff, D. R., Churchett, J. D., 1977: Prediction of flowering in sunflowers. Australian Journal of Experimental Agriculture and Animal Husbandry, 17, 475-481.

128

Gujja, B., Dalai, S., Shaik, H., Goud, V., 2009: Adapting to climate change in the Godavari River basin of India by restoring traditional water storage systems. Climate and Development, 1(3), 229-240

Haan, C. T., 2002: Statistical methods in Hydrology. 2nd ed., Iowa State University Press, Ames, Iowa, 378pp.

Han, P., Wang, P. X., Zhang, S. Y., Zhu, D. H., 2010: Drought forecasting based on the remote sensing data using ARIMA models. Mathematical and Computer Modelling, 51, 1398-1403.

Hargreaves, G. H., Samani, Z. A., 1985: Reference Crop Evapotranspiration from Temperature. Applied Engineering in Agriculture, 1(2): 96-99.

Hasan, Q., Bourque, C., Meng, F.-R., Richards, W., 2007: Spatial mapping of growing degree days: an application of MODIS-based surface temperatures and enhanced vegetation index. Journal of Applied Remote Sensing, 1, 013511.

Henriksen, H. J., Troldborg, L., Nyegaard, P., Sonnenborg, T. O., Refsgaard J. C., Madsen B., 2003: Methodology for Construction, Calibration and Validation of a National Hydrological Model for Denmark. Journal of Hydrology, 280, 52-71.

Hertig, E., Jacobeit, J., 2008: Assessments of Mediterranean precipitation changes for the 21st century using statistical downscaling techniques. International Journal of Climatology, 28(8), 1025-1045.

Hertig, E., Jacobeit, J., 2014: Considering observed and future nonstationarities in statistical downscaling of Mediterranean precipitation. Theoretical & Applied Climatology, 122(3), 667-683.

Hillel, D., Rosenzweig, C., 2002: Desertification in relation to climate variability and Change. Advances in Agriculture, 77, 1-38.

Holawe, F., Dutter, R., 1999: Geostatistical study of precipitation series in Austria: Time and space. Journal of Hydrology, 219, 70-82.

Holman, I. P., 2006: Climate change impacts on groundwater recharge uncertainty, shortcomings, and the way forward? Hydrogeology Journal, 14(5), 637–647.

Horton, R. E., 1931: The Field, Scope and Status of the Science of Hydrology. Transactions American Geophysical Union, 12, 189-202.

Hosking, J. R. M., 1981: Fractional differencing. Biometrika, 68(1), 165-176.

129

Immerwahr, J., 2004: Public Attitudes on Higher Education - A trend analysis, 1993 to 2003. Public Agenda, National Center for Public Policy and Higher Education, National Center Report #04-2.

IPCC, 2000: The IPCC Special Report on Emissions Scenarios (SRES). IPCC, Geneva, Switzerland, 20pp.

IPCC, 2001: Climate Change 2001 - Contribution of the Working Group 1 to the Third IPCC Assessment Report. In: Houghton, J. T., Ding, Y., Griggs, D. J., Noguer, M., van der Linden, P. J., Dai, X., Maskell, K., Johnson C. A., (eds.) Climate Change 2001: Synthesis Report. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

IPCC, 2007: Contribution of Working Group 1 to the Fourth IPCC Assessment Report. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K. B., Tignor, M., Miller, H. L., (eds.) Climate Change 2007: The Physical Science Basis. Cambridge. UK: Cambridge University Press, 996 p.

IPCC, 2013: Summary for Policymakers. In: Stocker, T. F., Qin, D., Plattner, G. K., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex., V., Midgley, P. M. (eds.) Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

IPCC, 2014: Climate Change 2014: Synthesis Report. An Assessment of Intergovernmental Panel on Climate Change. Geneva Switzerland, [accessed: 25/06/2016]

Irmak, S., Payero, J. O., Martin, D. L., Irmak, A., Howell, T. A., 2006: Sensitivity analyses and sensitivity coefficients of standardized daily ASCE Penman-Monteith equation. Journal of Irrigation and Drainage Engineering, 132(6), 564–578.

Jabloun, M., Sahli, A., 2008: Evaluation of FAO-56 methodology for estimating reference evapotranspiration using limited climatic data - Application to Tunisia: Agricultural Water Management, 95, 707-715.

Jacobs, J. M., Satti, S. R., 2001: Evaluation of reference evapotranspiration methodologies and AFSIRS crop water use simulation model, Final Report.. St. Johns River Water Management District, Palatka, FL.

Jaeger, E. B., Anders, I., Lüthi, D., Rockel, B., Schär, C., Seneviratne, S., 2008: Analysis of ERA40-driven CLM simulations for Europe. Meteorologische Zeitschrift, 17, 349-367.

130

Jefferies, R. A., Mackerron, D. K. L., 1987: Thermal time as a non-destructive method of estimating tuber initiation in potatoes. Journal of Agricultural Science, 108, 249-252.

Jensen, M. E., Haise, H. R., 1963: Estimating evapotranspiration from solar radiation. Journal of Irrigation and Drainage Engineering, 89, 15–41.

Johns, T. C. 2009a: ENSEMBLES STREAM2 METO-HCHADCM3C 20C3M run1, daily values. CERA database. World Data Center for Climate, Hamburg.

[accessed: 14/04/2016]

Johns, T. C. 2009b: ENSEMBLES STREAM2 METO-HCHADGEM2AO 20C3M run1, daily values. CERA database. World Data Center for Climate, Hamburg.

[accessed: 14/04/2016]

Jungclaus, J. H., Keenlyside, N., Botzet, M., Haak, H., Luo, J. J., Latif, M., Marotzke, J., Mikolajewicz, U., Roeckner, E., 2006: Ocean circulation and tropical variability in the coupled model ECHAM5/MPI-OM. Journal of Climate, 19, 3952-3972.

Kalogeropoulos, K., Chalkias, C., 2013: Modelling the impacts of climate change on surface runoff in small Mediterranean catchments: empirical evidence from Greece. Water and Environment Journal, 27, 505-513.

Karavitis, C. A., Alexandris, S., Tsesmelis, D. E., Athanasopoulos, G., 2011: Application of the Standardized Precipitation Index (SPI) in Greece. Water, 3, 787-805.

Kashaigili, J.J., Rajabu, K., Masolwa, P., 2009: Freshwater management and climate change adaptation: Experiences from the Great Ruaha River catchment in Tanzania. Climate and Development, 1(3), 220-228.

Kendall, M. G. 1938: A new measure of rank correlation. Biometrika, 30, 81-93.

Ketring, D. L., Wheless, T. G., 1989: Thermal time requirements for phenological development of peanut. Agronomy Journal, 81, 910- 917.

Kirby, E. J. M., 1995: Factors affecting rate of leaf emergence in barley and wheat. Crop Science, 35, 11-19.

Komuscu, A. U., 2001: An analysis of recent drought conditions in Turkey in relation to circulation patterns. Drought Network News, 13, 2-3.

Köse, B., 2014: Phenology and Ripening of Vitis vinifera L. and Vitis labrusca L. Varieties in the Maritime Climate of Samsun in Turkey's Black Sea Region. South African Journal of Enology and Viticulture, 35(1), 90-102. 131

Kostinakis, K., Xystrakis, F., Theodoropoulos, K., Stathis, D., Eleftheriadou, E., Matzarakis, A., 2011: Estimation of reference potential evapotranspiration with focus on vegetation science - the EmPEst software. Journal of Irrigation and Drainage Engineering, 137, 616-619.

Kotoulas, D., 2001: Hydrology and Hydraulics of the Natural Environment. Thessaloniki: Publication Division, A.U.Th., Thessaloniki, Greece. (In Greek)

Kotsopoulos, S., Babajimopoulos, C., 1997: Analytical estimation of modified Penman equation parameters, Journal of Irrigation and Drainage Engineering, 123(4), 253-256.

Kotsopoulos, S., Nastos, P., Lazogiannis, K., Poulos, S., Ghionis, G., Alexiou, I., Panagopoulos, A., Farsirotou, E., Alamanis, N., 2015: Evaporation, evapotranspiration and crop water requirements under present and future climate conditions at Pinios delta plain. In: Proceedings of the 14th International Conference on Environmental Science and Technology. Rhodes, Greece, 3-5 September 2015, CEST2015_00647.

Koufos, G. C., Mavromatis, T., Koundouras, S., Fyllas, N. M., Jones, G. V., 2014: Temporal evolution of bioclimatic indices in the main winegrape regions of Greece. In: Kanakidou, M., Mihalopoulos, N., Nastos, P. T., (eds.) COMECAP e-book of contributions. In: Proceedings of the 12th International Conference of Meteorology, Climatology and Physics of the Atmosphere. Heraklion, Greece, 28-31 May 2014, vol. 2, 45-49.

Koutsoyiannis, D., Efstratiadis, A., Georgakakos, K., 2008: Uncertainty Assessment of Future Hydroclimatic Predictions: A Comparison of Probabilistic and Scenario-Based Approaches. Journal of Hydrometeorology, 8, 261-281.

Lenhart, T., Eckhardt, K., Fohrer, N., Frede, H. G., 2002: Comparison of two different approaches of sensitivity analysis. Physics and Chemistry of the Earth, 27, 645-654.

Ljung, G. M., Box, G. E. P., 1978: On a measure of lack of fit in time series models. Biometrika, 65(2), 297-303.

Loukas, A., Vasiliades, L., Tzabiras, J., 2008: Climate change effects on drought severity. Advances in Geosciences, 17, 23-29.

Lu, J. B., Sun, G., McNulty, S. G., Amatya, D. M., 2005: A comparison of six potential evapotranspiration methods for regional use in the southeastern United States. American Water Resources Association, 41(3), 621–633.

L'vovich, M. I., 1979: World Water Resources and their future. (English translation from Russian), edited by R. L. Nace. American Geophysical Union, Washington DC.

132

Maliva, R. G., Missimer, T. M., 2012: Arid Lands Water Evaluation and Management. Springer-Verlag Berlin Heidelberg, p. 1076.

Mann, H. B., 1945: Nonparametric tests against trend. Econometrica, 13, 245-259.

Maris, F., Paparrizos, S., Karatzios, G., 2014: Hydrogeoinformatics - Modeling and Information Systems for Water Resources Management. Disigma Publications, Thessaloniki, Greece, p. 376.

Maris, F., Kitikidou, K., Angelidis, P., Potouridis, S., 2013: Kriging interpolation method for estimation of continuous spatial distribution of precipitation in Cyprus. British Journal of Applied Science & Technology, 3(4), 1286-1300.

Maris, F., Kitikidou, K., Paparrizos, S., Karagiorgos, K., Potouridis, S., Fuchs, S., 2015: Regional Hazard Analysis for Use in Vulnerability and Risk Assessment. Quaestiones Geographicae, 34(3), 77-84.

Martinez-Cob, A., 1996: Multivariate geostatistical analysis of evapotranspiration and precipitation in mountainous terrain. Journal of Hydrology, 174, 19-35.

Masle, J., Doussinalut, G., Farquhar, G. D., Sun, B., 1989: Foliar stage in wheat correlates better to photothermal time than to thermal time. Plant Cell Environment, 12, 235-247.

Masoni, A., Ercoli, L., Massantini, F., 1990: Relationship between number of days, growing degree days and photothermal units and growth in wheat (Triticurn uestiuum L.) according to seeding time. Annals of Agricultural and Environmental Medicine, 120, 41 -51.

Mathan, K. K., 1989: Influence of accumulated heat units and sunshine hours on the growth and yield of sorghum (var. Co 25). Journal of Agronomy and Crop Science, 163, 196- 200.

MATLAB and Statistics Toolbox, 2000: Release 2014a. The Mathworks, Inc., Natick, Massachusetts, United States.

Matzarakis A., 2006: The climate of Evros. Freiburg, Germany, 38p. ISBN-10:3-00-020071- 1.

Matzarakis, A., Ivanova, D., Balafoutis, C., Makrogiannis, T., 2007: Climatology of growing degree days in Greece. Climate Research, 34, 233-240.

McCuen, H. R., 1974: A sensitivity and error analysis of procedures used for estimating evapotranspiration. Water Resources Bulletin, 10(3), 486-498.

133

McKee, T. B., Doesken, N. J., Kleist, J., 1993: The relationship of drought frequency and duration to time scales. In: Proceedings of the 8th conference on Applied Climatology. Anaheim, CA, U.S.A., pp. 179-184.

McKee, T. B., Doesken, N. J., Kleist, J., 1995: Drought monitoring with multiple scales. In: Proceedings of the 9th conference on Applied Climatology. Boston, MA, U.S.A., pp 233-236.

McMaster, G. S., Smika, D. E., 1988: Estimation and evaluation of winter wheat phenology in the central Great plains. Agricultural and Forest Meteorology, 43, 1-18.

McMaster, G. S., Wilhelm, W. W., 1997: Growing degree-days: one equation, two interpretations. Agricultural and Forest Meteorology, 87, 291-300.

McMaster, G. S., Wilhelm, W. W., Morgan, J. A., 1992: Simulating winter wheat shoot apex phenology. Journal of Agricultural Science, 119, 1-12.

Meehl G. A., Arblaster J.M., Tebaldi C., 2005: Understanding future patterns of increased precipitation intensity in climate model simulations. Geophysical Research letters, 32, L18719.

Metaxas, D., Philandras, C., Nastos, P., Repapis, C., 1999: Variability of precipitation pattern in Greece during the year. Fresenius Environmental Bulletin, 8, 1–6.

Michaelides, S., Pashiardis, S., 2008: Monitoring Drought in Cyprus during the 2007-2008 Hydro-meteorological Year by using the Standardized Precipitation Index. European Water, 23/24, 123-131.

Miley, W. N., Oosterhuis, D. M., 1990: Nitrogen Nutrition of Cotton: Practical Issues. In: Proceedings of 1st Annual Workshop for Practicing Agronomists. Arizona, U.S.A., February 7, 1990. American Society of Agronomy, Inc..

Miller, P., Lanier, W., Brandt, S., 2001. Using Growing Degree Days to Predict Plant Stages. Ag/Extension Communications Coordinator, Communications Services, Montana State University-Bozeman, Bozeman, MO.

Milly, P. C. D., Dunne, K. A., Vecchia, A. V., 2005: Global pattern of trends in streamflow and water availability in a changing climate, Nature, 438, 347-350.

Mimikou, M. A., 2005: Water Resources in Greece: Present and future. Global NEST Journal, 7(3), 313-322.

Mimikou, M. A., Baltas, E., Varanou, E., Pantazis, K., 2000: Regional Impacts of Climate Change on Water Resources Quality and Quantity Indicators. Journal of Hydrology, 234, 95-109.

134

Moench, M., Dixit, A., Janakarajan, S., Rathore, M. S., Mudrakartha, S., 2003: The Fluid Mosaic: Water Governance in the Context of Variability, Uncertainty and Change – A Synthesis Paper. Nepal Water Conservation Foundation, Kathmandu, 71 pp.

Mora, D. E., Campozano, L., Cisneros, F., Wyseure, G., Willems, P., 2014: Climate changes of hydrometeorological and hydrological extremes in the Paute basin, Ecuadorean Andes. Hydrological Earth Systems Sciences, 18, 631-648.

Morrison, W., Andresen, J., Szendrei, Z., 2014: The development of the asparagus miner (Ophiomyia simplex Loew; Diptera: Agromyzidae) in temperate zones: a degree-day model. Pest Management Science, 70(7), 1105-1113.

Muthers, S., Matzarakis, A., 2010: Use of beanplots in applied climatology - A comparison with boxplots. Meteorologische Zeitschrift, 19(6), 639-642.

Myers, D., 1982: Matrix formulation of co-kriging. Journal of Mathematical Geology, 14(3), 249-257.

Narwal, S. S., Poonia, S., Singh, G., Malik, D. S., 1986: Influence of sowing dates on the growing degree days and phenology of winter maize (Zea mays L.). Agricultural and Forest Meteorology, 38, 47-57.

Nastos, P., Zerefos, C., 2007: On extreme daily precipitation totals at Athens, Greece. Advances in Geosciences, 10, 59–66.

Nastos, P., Zerefos, C., 2008: Decadal changes in extreme daily precipitation in Greece. Advances in Geosciences, 16, 55-62.

Nastos, P., Zerefos, C., 2009: Spatial and temporal variability of consecutive dry and wet days in Greece. Atmospheric Research, 94, 616-628.

Nastos, P., Politi, N., Kapsomenakis, J., 2013a: Spatial and temporal variability of the Aridity Index in Greece. Atmospheric Research, 119, 140-152.

Nastos, P., Kapsomenakis, J., Douvis, K., 2013b: Analysis of precipitation extremes based on satellite and high-resolution gridded data set over Mediterranean basin. Atmospheric Research, 131, 46-59.

OECD, 2014: Climate Change, Water and Agriculture: Towards Resilient Systems. OECD Studies on Water, OECD Publishing, 104pp.

Oliver, M. A., Webster, R., 1990: Kriging: a method of interpolation for geographical information systems. International Journal of Geographical Information Systems, 4(3), 313-332.

135

Open Geospatial Consortium, 2016: OGC network Common Data Form (netCDF) standards suite. < http://www.opengeospatial.org/standards/netcdf> [accessed: 18/04/2016]

Orlandi, F., Vazquez, L. M., Ruga, L., Bonofiglio, T., Fornaciari, M., Garcia-Mozo, H., Dominguez, E., Romano, B., Galan, C., 2005: Bioclimatic requirements for olive flowering in two Mediterranean regions located at the same latitude (Andalucía, Spain, and Sicily, Italy). Annals of Agricultural and Environmental Medicine, 12, 47-52.

Orlandi, F., Garcia-Mozo, H., Dhiab, B., Galan, C., Msallem, M., Fornaciari, M., 2014: Olive tree phenology and climate variations in the Mediterranean area over the last two decades. Theoretical & Applied Climatology, 115, 207-218

Oudin, L., Hervieu, F., Michel, C., Perrin, C., Andréassian, V., Anctil, F., Loumagne, C., 2005: Which potential evapotranspiration input for a lumped rainfall-runoff model? Part 2 - Towards a simple and efficient potential evapotranspiration model for rainfall-runoff modeling. Journal of Hydrology, 303(01-avr), 290-306.

Paparrizos, S., 2012: Study of the torrential environment of the Sperchios River with the integrated, hydrological, physically-based MIKE-SHE model, using GIS. MSc. Thesis, Democritus University of Thrace, Orestiada, Greece, 180pp. (In Greek)

Paparrizos, S., Chatziminiadis, A. M., 2010: Study of the torrential environment of Sperchios River. BSc. Thesis, Democritus University of Thrace, Orestiada, Greece, 166pp. (In Greek)

Paparrizos, S., Matzarakis, A., 2016: Assessment of future climate change impacts on the hydrological regime of selected Greek areas with different climate conditions. Hydrology Research, DOI 10.2166/nh.2016.018

Paparrizos, S., Maris, F., Matzarakis, A., 2014: Estimation and Comparison of Potential Evapotranspiration based on daily and monthly data from Sperchios River valley in Central Greece. Global NEST Journal, 16(2), 204-217.

Paparrizos, S., Maris, F., Kitikidou, K., Anastasiou, Th., Potouridis, S., 2015: Comparative analysis of soil erosion quantizations within GIS environment: an application on Sperchios River basin in Central Greece. International Journal of River Basin Management, 13(4), 475-486.

Paparrizos, S., Maris, F., Papageorgiou, O., Karagiorgos, K., Fuchs, S., Matzarakis, A., 2015: Sensitivity analysis of various Potential Evapotranspiration formulas for Crete Island in Greece. In: Proceedings of Geophysical Research Abstracts, 17, EGU2015- 8556-1, Vienna, Austria.

136

Paparrizos, S., Maris, F., Matzarakis, A., 2016a: Integrated analysis of present and future responses of precipitation over selected Greek areas with different climate conditions. Atmospheric Research, 169, 199-208.

Paparrizos, S., Maris, F., Matzarakis, A., 2016b: Integrated analysis and mapping of aridity over Greek areas with different climate conditions. Global NEST Journal, 18(1), 131- 145.

Paparrizos, S., Maris, F., Matzarakis, A., 2016c: Sensitivity analysis and comparison of various potential evapotranspiration formulae for selected Greek areas with different climate conditions. Theoretical & Applied Climatology, DOI: 10.1007/s00704-015- 1728-z

Paparrizos, S., Maris, F., Matzarakis, A., 2016d: Growing Degree Days and precipitation as assessment factors for future responses in Evros region, Greece. In: Proceedings of the 3rd Conference on Civil Protection (SafeEvros 2016). Alexandroupoli, Greece, 22- 25 June 2016, 9.

Park, C., 2001: The Environment: Principles and Applications. 2nd edition, Routledge, NY, 704 pp.

Parry, M.L., Canziani, O. F., Palutikof, J. P., et al., 2007: Technical Summary. Climate Change 2007: Impacts, Adaptation and Vulnerability. In: Parry, M. L., Canziani, O. F., Palutikof, J. P., van der Linden, P. J., Hanson, C. E., (eds.) Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, 23-78.

Pashiardis, S., Michaelides, S., 2008: Implementation of the Standardized Precipitation Index (SPI) for Regional Drought Assessment: A case study for Cyprus. European Water, 23/24, 57-65.

Pearson, K., 1900: Mathematical contributions to the theory of evolution, VII: On the correlation of characters not quantitatively measurable. Philosophical Transactions of the Royal Society of London, A195, 1-147.

Peel, M. C., Finlayson, B. L., McMahon, T. A., 2007: Updated world map of the Köppen- Geiger climate classification. Hydrology and Earth Systems Sciences, 11, 1633–1644.

Perlman, H., Makropoulos, C., Koutsoyiannis, D., 2005: The water cycle. United States Geological Syrvey.

Perry, K. B., Wehner, T. C., Johnson, G. L., 1986: Comparison of 14 methods to determine heat unit requirements for cucumber harvest. HortScience, 21, 419-423.

137

Piccarreta, M., Capolongo, D., Boenzi, F., 2004: Trend analysis of precipitation and drought in Basilicata from 1923 to 2000 within a Southern Italy context. International Journal of Climatology, 24, 907–922.

Piervitali, E., Colasino, M., Conte, M., 1997: Signals of climatic change in the Central- Western Mediterranean basin. Theoretical & Applied Climatology, 58, 211–219.

Pimentel, D., Houser, J., Preiss, E., White, O., Fang, H., Mesnick, L., Barsky, T., Tariche, S., Schreck, J., Alpert, S., 1997: Water Resources: Agriculture, the Environment, and Society. BioScience, 47(2), 97-106.

Priestley, C. H., Taylor, R. J., 1972: On the assessment of surface heat flux and evaporation using large-scale parameters. Monthly Weather Review, 100, 81-92.

Raes, D., Steduto, P., Hsiao, T. C., Fereres, E., 2010: ANNEXES Reference Manual. AquaCrop. 50 p.

Rakhecha, P. R., Singh, V. P., 2009: Applied Hydrometeorology. Springer, Dordrecht, The Netherlands with Capital Publishing Company, New Delhi, India, 384pp.

Repapis, C., 1986: Temporal fluctuations of precipitation in Greece. Rivista di Meteorologia Aeronautica, XLVI(1–2), 19–25.

Roeckner, E., Brokopf, R., Esch, M., Giorgetta, M., Hagemann, S., Kornblueh, L., Manzini, E., Schlese, U., Schulzweida, U., 2006: Sensitivity of simulated climate to horizontal and vertical resolution in the ECHAM5 Atmosphere Model. Journal of Climate, 19, 3771-3791.

Rosenberry, D. O., Stannard, D. I., Winter, T. C., Martinez, M. L., 2004: Comparison of 13 equations for determining evapotranspiration from a prairie wetland, Cottonwood Lake Area, North Dakota, USA. Wetlands Ecology and Management, 24(3), 483–497.

Russelle, M. P., Wilhelm, W. W., Olson, R. A., Power, J. F., 1984: Growth analysis based on degree days. Crop Science, 24, 28-32.

Ruth, M., Davidsdottir, B., Amato, A., 2004: Climate change policies and capital vintage effects: the case of U.S. pulp and paper, iron and steel, and ethylene. Journal of Environmental Management, 70, 235–252.

Salmi, T., Määttä, A., Anttila, P., Ruoho-Airola, T., Amnell, T. 2002: Detecting trends of annual values of atmospheric pollutants by the Mann-Kendall test and Sen’s slope estimates – the Excel template application MAKESENS. Publications on air quality, Finnish Meteorological Institute, Air Quality Research, No. 31, Helsinki.

138

Sanchez-Gomez, E., Somot, S., Mariotti, A., 2009: Future changes in the Mediterranean water budget projected by an ensemble of regional climate models. Geophysical Research Letters, 36, L21401.

Saravi, M., Safdari, A., Malekian, A., 2009: Intensity-Duration-Frequency and spatial analysis of droughts using the Standardized Precipitation Index. Hydrology and Earth System Sciences Discussions, 6, 1347-1383.

Saxton, K. E., 1975: Sensitivity analysis of the combination evapotranspiration equation. Agricultural Meteorology, 15, 343-353.

Schubert, S., Henderson-Sellers, A., 1997: A statistical model to downscale local daily temperature extremes from synoptic-scale atmospheric circulation patterns in the Australian region. Climate Dynamics, 13, 223–234.

Schmidt, O., 1949: A theory of Earth's origin: Four lectures. Moscow: Izdatelstvo Akad. Nauk SSSR. (In Russian)

Schönwiese, C. D., Rapp, J., Fuchs, T., Denhard, M., 1994: Observed climate trends in Europe 1891–1990. Meteorologische Zeischrift, 38, 51–63.

Scudder, T., 2005: The Future of Large Dams. Earthscan, London, 408 pp

Snedecor, G. W., Cochran, W. G., 1989: Statistical methods. 8th ed., Iowa State University Press, Ames, Iowa, 803 pp.

Stern, N., 2007: The Economics of Climate Change: The Stern Review. Cambridge University Press, Cambridge, 692 pp.

Swanson, S. P., Wilhelm, W. W., 1996. Planting date and residue rate effects on growth, partitioning, and yield of corn. Agronomy Journal, 88, 205-210.

Tigkas, D., Vangelis, H., Tsakiris, G., 2012: Drought and climatic change impact on streamflow in small watersheds. Science of the Total Environment, 440, 33-41.

Tigkas, D., Vangelis, H., Tsakiris, G., 2015: DrinC: a software for drought analysis based on drought indices. Earth Science Informatics, 8, 697-709.

Timbal, B., Dufour, A., McAvaney, B., 2003: An estimate of future climate change for western France using a statistical downscaling technique. Climate Dynamics, 20, 807– 823.

Tolika, K., Anagnostopoulou, C., Maheras, P., Vafiadis, M., 2008: Simulation of future changes in extreme rainfall and temperature conditions over the Greek area: A comparison of two statistical downscaling approaches. Global and Planetary Change, 63, 132-151.

139

Tollenaar, M., Daynard, T. B., Hunter, R. B., 1979: Effect of temperature on rate of leaf appearance and flowering date in maize. Crop Science, 19, 363-366.

Tsakiris, G., Vangelis, H., 2004. Towards a Drought Watch System based on spatial SPI. Water Resources Management, 18(1), 1-12.

Tsakiris, G., Vangelis, H., 2005: Establishing a Drought Index Incorporating Evapotranspiration. European Water, 9/10, 3-11.

Türkes, M., 2003: Spatial and temporal variations in precipitation and aridity index series in Turkey. In: Bölle, H. J., (eds.) Mediterranean climate. Variability and trends. Springer, pp 181–213.

United Nations Environment Programme (UNEP), 1992: World Atlas of Desertification. Edward Arnold, London, UK.

UNEP/GRID-Arendal, 2002: Vital Climate Graphics. United Nations Environment Programme. [accessed: 23/05/2015]

Uyarra, M., Cote, I., Gill, J., Tinch, R., Viner, D., Watkinson, A. L., 2005: Island-specific preferences of tourists for environmental features: implications of climate change for tourism-dependent states. Environmental Conservation, 32(1), 11–19.

Valiantzas, J., 2006: Simplified versions for the Penman evaporation equation using routine weather data. Journal of Hydrology, 331, 690-702.

Van Griensven, A., Francos, A., Bauwens, W., 2002: Sensitivity analysis and auto- calibration of an integral dynamic model for river water quality. Water Science and Technology, 45(9), 325-332.

Van Griensven, A., Meixner, T., Grunwald, S., Bishop, T., Dilluzio, M., Srinivasan, R., 2006: A global sensitivity analysis tool for the parameters of multivariable catchment models. Journal of Hydrology, 324(1-4), 10-23.

Vangelis, H., Tigkas, D., Tsakiris, G., 2013: The effect of PET method on Reconnaissance Drought Index (RDI) calculation. Journal of Arid Environments, 88, 130-140

Vasiliades, L., Loukas, A., Patsonas, G., 2009: Evaluation of a statistical downscaling procedure for the estimation of climate change impacts on droughts. Natural Hazards and Earth System Sciences, 9, 879-894.

Vernadsky, V. I., 1965: The Chemical Structure of the Earth’s Biosphere and Its Surroundings. Moscow: Science, 375pp.

140

Voudouris, K., Mavromatis, T., Krinis, P., 2012: Assessing runoff in future climate conditions in Messara valley in Crete with a rainfall-runoff model. Meteorological Applications, 19, 473-483.

Vörösmarty, C.J., Green P., Salisbury J., Lammers, R. B., 2000: Global water resources: vulnerability from climate change and population growth. Science, 289, 284–288.

Vörösmarty, C.J., McIntyre, P. B., Gessner, M. O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S. E., Sullivan, C.A., Reidy Liermann, C., Davies, P. M., 2010: Global threats to human water security and river biodiversity. Nature, 467, 555-561.

Vroxidou, A. E., Grillakis, M. G., Tsanis, K., 2013: Drought Assessment based on a Multi- Model precipitation projections for the Island of Crete. Earth Science and Climatic Change, 4, 158.

Wang, J. Y., 1960: A critique of the heat unit approach to plant response studies. Ecology, 41, 785-790.

Wang, G., 2005: Agricultural drought in a future climate: results from 15 global climate models participating in the IPCC 4th assessment. Climate Dynamics, 25(7–8), 739– 753.

Wang, G., Minnis, R. B., Belant, J.L., Wax C. L., 2010: Dry weather induces outbreaks of human West Nile virus infections. BMC Infectious Disease Journal, 10(38), 1–7.

Wendling, U., Müller, J., Schwede, K., 1984: Ergebnisse von Verdunstungsmessungen über Gras mit einem Off-line-Datenerfassungssystem. Zeitschrift für Meteorologie, 34(2), 190–202. (In German)

Wilhelm, W. W., McMaster, G. S., 1995: The importance of the phyllochron in studying the development of grasses. Crop Science, 35, 1-3.

Wilhelm, W. W., Bouzerzour, H., Power, J. F., 1989: Soil disturbance-residue management effect on winter wheat growth and yield. Agronomy Journal, 81, 581-588.

Wilhelm, W. W., Schepers, L. N., Mielke, J. S., Doran, J. W., Ellis, J. R., Stroup, W. W., 1987: Dryland maize development and yield resulting from tillage and nitrogen fertilization practices. Soil & Tillage Research, 10, 167-179.

Winchell, M., Srinivasan, R., Di Luzio M., Arnold, J. G., 2013: ArcSWAT Interface for SWAT 2012: User's Guide. Texas Agricultural Experiment Station (Texas) and USDA Agricultural Research Service (Texas), Temple, TX, pp. 464.

141

Winter, T. C., Rosenberry, D. O., Sturrock, A. M., 1995: Evaluation of 11 equations for determining evaporation for a small lake in the north central United States. Water Resources Research, 31, 983–993.

Xu, C. Y., Singh, V. P., 2002: Cross comparison of empirical equations for calculating potential evapotranspiration with data from Switzerland. Water Resources Management, 16(3), 197–219.

Xystrakis, F., Matzarakis, A., 2011: Evaluation of 13 empirical reference potential evapotranspiration equations on the island of Crete is southern Greece. Journal of Irrigation and Drainage Engineering, 137, 211-222.

Yao, H., 2009: Long-term study of lake evaporation and evaluation of seven estimation methods: results from Dickie Lake, South-Central Ontario, Canada. Journal of Water Resources and Protection, 1(2), 59-77.

Yin, J. H., 2005: A consistent poleward shift of the storm tracks in simulations of 21st century climate. Geophysical Research letters, 32, L18701.

Yu, X., Jiang, L., Li, L., Wang, J., Wang, L., Lei, G., Pittock, J., 2009: Freshwater management and climate change adaptation: Experiences from the Central Yangtze in China. Climate and Development, 1(3), 241-248.

Yurekli, K., Kurunc, K., Ozturk, F., 2005: Application of linear stochastic models to monthly flow data of Kelkit stream. Ecological Modeling, 183, 67-75.

142

List of Figures

Figure 2.01. The hydrological cycle...... 25 Figure 2.02. Supply-demand water situation for the 14 water districts in Greece...... 35 Figure 3.01. Location and characteristics of the study areas...... 38 Figure 3.02. Meteorological Stations - Ardas River basin...... 39 Figure 3.03. Land Use - Ardas River basin...... 39 Figure 3.04. Meteorological Stations - Sperchios River basin...... 40 Figure 3.05. Land Use - Sperchios River basin...... 40 Figure 3.06. Meteorological Stations - Geropotamos River basin...... 41 Figure 3.07. Land Use - Geropotamos River basin...... 41 Figure 3.08. Köppen-Geiger classification - Ardas River basin...... 43 Figure 3.09. Köppen-Geiger classification - Sperchios River basin...... 43 Figure 3.10. Köppen-Geiger classification - Geropotamos River basin...... 43 Figure 3.11. Dynamical and statistical downscaling procedure...... 50 Figure 4.01. Beanplots of the mean daily PET (mm) for Orestiada station in Ardas River basin, based on the PET formulae examined in Table 3.08...... 63 Figure 4.02. Beanplots of the mean daily PET (mm) for Metaxades station in Ardas River basin, based on the PET formulae examined in Table 3.08...... 63 Figure 4.03. Beanplots of the mean daily PET (mm) for Lamia station in Sperchios River basin, based on the PET formulae examined in Table 3.08...... 64 Figure 4.04. Beanplots of the mean daily PET (mm) for Tympaki station in Geropotamos River basin, based on the PET formulae examined in Table 3.08...... 64 Figure 4.05. Beanplots of the mean daily PET (mm) for Zaros station in Geropotamos River basin, based on the PET formulae examined in Table 3.08...... 64 Figure 4.06. Precipitation analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 74 Figure 4.07. Precipitation analysis of the Sperchios River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 75 Figure 4.08. Precipitation analysis of the Geropotamos River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 76 Figure 4.09. Observed and simulated discharge (m3/s) for examined reference periods in the study areas...... 78

143

Figure 4.10. Schematic representation of trend analysis results regarding the variation of runoff in the upcoming years in the study areas...... 81 Figure 4.11. Growing Degree Days analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 86 Figure 4.12. Growing Degree Days analysis of the Sperchios River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 87 Figure 4.13. Growing Degree Days analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 88 Figure 4.14. Aridity Index analysis maps for Ardas River basin a)Total Analysis b) Spring c) Summer d) Autumn d) Winter...... 91 Figure 4.15. Aridity Index analysis maps for Sperchios River basin a)Total Analysis b) Spring c) Summer d) Autumn d) Winter...... 92 Figure 4.16. Aridity Index analysis maps for Geropotamos River basin a)Total Analysis b) Spring c) Summer d) Autumn d) Winter...... 93 Figure 4.17. Drought analysis of the Ardas River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 98 Figure 4.18. Drought analysis of the Sperchios River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 99 Figure 4.19. Drought analysis of the Geropotamos River basin, based on the A1B and B1 scenarios for the future periods of 2021-2050 and 2071-2100...... 100 Figure 5.01. Synthesis report graph of the results from the current study...... 118

144

List of Tables

Table 2.01. Global Water Balance...... 24 Table 2.02. Distribution of freshwater on Earth ...... 24 Table 2.03. Supply-demand water situation for the 14 water districts in Greece...... 34 Table 3.01. Characteristics of the study areas...... 37 Table 3.02. Meteorological stations and characteristics - Ardas River basin...... 39 Table 3.03. Meteorological stations and characteristics - Sperchios River basin...... 40 Table 3.04. Meteorological stations and characteristics - Geropotamos River basin...... 41 Table 3.05. Driving GCMs, Institutes, RCMs, and relevant references of the used ENSEMBLES simulations...... 44 Table 3.06. IPCC emission scenarios ...... 45 Table 3.07. Categorization of the output p-values at 95% significance level (p≤ 0.05)...... 49 Table 3.08. Potential Evapotranspiration formulae...... 53 Table 3.09. Input data requirements of used PET formulae...... 54 Table 3.10. Classification of Aridity Index categories...... 56 Table 3.11. Classification scale for the SPI values...... 57 Table 3.12. Main cultivations and their characteristics...... 62 Table 4.01. Sensitivity analysis results (%) - Orestiada...... 66 Table 4.02. Sensitivity analysis results (%) - Metaxades ...... 66 Table 4.03. Sensitivity analysis results (%) - Lamia...... 66 Table 4.04. Sensitivity analysis results (%) - Tympaki...... 66 Table 4.05. Sensitivity analysis results (%) - Zaros station...... 67 Table 4.06. Ranking of PET formulae according to the average results of each station...... 68 Table 4.07. Precipitation analysis results of present and future precipitation conditions in the study areas...... 70 Table 4.08. Trend analysis results of present and future precipitation conditions in the study areas...... 71 Table 4.09. Seasonal analysis results of present and future precipitation conditions in the study areas...... 72 Table 4.10. ArcSWAT simulation results...... 79 Table 4.11. Trend analysis results of present and future runoff conditions in the study areas...... 81

145

Table 4.12. Average irrigation water withdrawal in Greece and agricultural water withdrawal Versus total water withdrawal...... 82 Table 4.13. Classification of the study areas regarding the severity of future climate change...... 82 Table 4.14. Reference (observed) and future (RCMs) simulations periods' data validation..83 Table 4.15. Accumulated mean yearly GDD (oC) for reference, 2021-2050 and 2071-2100 periods...... 84 Table 4.16. Accumulated mean GDD (oC) starting from April for reference, 2021-2050 and 2071-2100 periods...... 84 Table 4.17. Aridity index analysis results for the selected Greek areas...... 89 Table 4.18. Reference (observed) and future (RCMs simulations) periods' data validation...95 Table 4.19. Drought analysis results occurred by the average means of the examined RCMs...... 95 Table 4.20. Trend analysis results of present and future drought conditions in the study areas...... 96

146

List of Symbols and Abbreviations

ACF Auto -correlation function AET Actual Evapotranspiration AI Aridity Index AR4 IPCC 4th Assessment Report ArcSWAT ArcGIS extension - Soil and Water Assessment Tool ARIMA Auto-regressive integrated moving average ASCE American Society of Civil Engineers Bjerknes Centre for Climate Research - Bergen Climate Model - BCCR-BCM2 Version 2 bo Constant coefficient from the multi-linear regression equation coefficients obtained for each independent factor from the multi-linear b1...bn regression equation Denominator constant that changes with reference type and takes the -1 -1 Cd values of 0.34 (s m ) and 0.38 (s m ) for short and tall crops, respectively

Numerator constant that changes with reference type. Cn takes the Cn value 900 (K mm s3 Mg-1) for short and tall crops, respectively Centre National de Recherches Meteorologiques global climate model CNRM-CM3 version 3 Centre National de Recherches Meteorologiques regional climate CNRM-RM4.5 model version 4.5

CO2 Carbon Dioxide d Number of lagged forecast errors in the prediction equation DistWat Distance from a body of water or a lake (km) DL Day Length (hours) ea Mean actual vapour pressure at 1.5-2.5 m height (kPa) EC European Commission European Centre Hamburg Model / Max Plank Institute - Ocean ECHAM/MPI-OM1 Model version 1

147

ECT Eddy-Covariance-Technique EF Coefficient of Efficiency (Nash-Sutcliffe) es Mean saturation vapour pressure at 1.5-2.5 m height (kPa) ET Evapotranspiration

ETref Reference Evapotranspiration Swiss Federal Institute of Technology in Zurich - Community Land ETHZ-CLM Model FAO Food and Agriculture Organization of the United Nations f(t) Continuous monotonic increasing or decreasing function of time G Soil heat flux density at the crop surface (MJ m-2) GCM Global Circulation Model GDD Growing Degree Days GIS Geographic Information Systems h Elevation (m) h Hours HADCM3C Hadley Centre Coupled Model version 3C HADGEM2 Hadley Centre Global Environment Model version 2 HadCMeQ0 Hadley Centre Coupled Model with normal sensitivity HIRHAM5 High Resolution Limited Area Model version 5 HNMS Hellenic National Meteorological Service I Inflow volume of water during a given time period IPCC Intergovernmental Panel on Climate Change

Ksp New coefficient (Sensitivity analysis) o Lat. (also: Xi, xi) Geographic Latitude ( ) o Long. (also: Yi, yi) Geographic Longtitude ( ) M Modelled value (Sensitivity analysis) Regional Atmospheric Climate Model High Resolution Limited Area METNO_HIRHAM Model METO-HC-HadCM3C Met Office Hadley Centre - Hadley Centre Coupled Model version 3C Met Office Hadley Centre - Hadley Centre Global Environment Model METO-HC-HadGEM2 version 2 n Number of annual values in the studied data series (Trend analysis) n Sample size (RMSE) N Sample size during trend analysis NAO North Atlantic Oscillation NAOI North Atlantic Oscillation Index NetCDF Network Common Data Form File

148

O Outflow volume of water during a given time period

Oi Values of observed (historical) discharge (EF equation) Ō Average of the historical discharge OECD Organisation for Economic Co-operation and Development p P-value (multi-linear regression output) p Examined independent variable or parameter (Sensitivity analysis) p Number of autoregressive terms (ARIMA)

Pi Values of Discharge predicted by the model (EF equation) PACF Partial Auto-correlation function PET Potential Evapotranspiration PREC Precipitation (mm) PSDI Palmer Drought Severity Index q Number of tied groups (Trend analysis) q Number of non-seasonal differences needed for stationarity (ARIMA) Q Discharge (m3/s) r Correlation coefficient between two samples during trend analysis r The radius of Earth -2 Ra Total solar radiation reaching the atmosphere's surface (MJ m ) RCM Regional Climate Model RCP Representative Concentration Pathways RDI Reconnaissance Drought Index REMO Regional Climate Model REMO RH Relative Humidity (%) RMSE Root Mean Square Error -2 Rn Total net solar radiation (MJ m ) -2 Rs Total global solar radiation (MJ m ) S S test (Trend analysis) slp Slope (%) SPI Standardized Precipitation Index SRES Special Report on Emission Scenarios std Standard Deviation SVD Saturated vapour density at mean air temperature (g m-3) SW Amount of water stored in the soil profile of the watershed (mm) SWAT Soil and Water Assessment Tool th tp number of data values in the p group T Air Temperature (oC)

Tbase Temperature below which the process of growth does into progress

149

(oC) o Tdew Dew point temperature ( C) o Tmean Mean air temperature ( C) o Tmax Maximum air temperature ( C) o Tmin Minimum air temperature ( C) -1 U2 Mean wind speed at 2 m height (m s ) UNEP United Nations Environmental Program U.S. United States of America VAR Variance VPD Vapor pressure deficit WS Wind Speed (m s-1) xi (Trend analysis) Data values of a time series that assume to obey the model

Xobs Observed values (RMSE)

Xmodel Model values (RMSE) Z Z value (Trend Analysis)

Zi, zi Altitude (m) α Proportionally coefficient (unitless - set as 1.2 in the current study)

αPT Priestley-Taylor coefficient (unitless - set as 1.26 in the current study) γ Psychrometric constant (kPa oC -1) Slope of saturation vapour pressure curve at air temperature T (kPa Δ oC -1) ΔS Change in water storage (mm) Residual that can be assummed to be from the same distribution with εi zero mean λ Latent heat of vaporization (MJ kg-1) π The ratio of a circle's circumference to its diameter (≈3.14) ρ Population correlation coefficient

σp Standard deviation of the meteorological variable's data series

150

Appendix

Appendix A1. Indicative 'read' of the ENSEMBLES plain text data from netCDF File in MATLAB...... 152 Appendix A2. Developed Script in MATLAB for exporting 'netCDF' files that contain the future simulations from the ENSEMBLES Project...... 156 Paper 1...... 157 Paper 2...... 175 Paper 3...... 187 Paper 4...... 217 Paper 5...... 243 Paper 6...... 261 Paper 7...... 287

151

Appendix A1. Indicative 'read' of the ENSEMBLES plain text data from netCDF File in MATLAB

% Command: ncdisp('BCM2_SRA1B_1_MM_tas.nc_1-1200');

Format: classic Global Attributes: title = 'BCCR model output prepared for IPCC Fourth Assessment Report 720 ppm stabilization experiment (SRES A1B)' description = 'CONTROL RUN ARPEGE - MICOM' north_pole = [0 90] source = 'BCM2.0, (2005): atmosphere: ARPEGE V3 T63L31 (cy 22b+); ocean : NERSC-MICOM V1.0 (based on MICOM V2.8), 35 vertical layers, 1.5(E-W) X 0.5(N-S) deg. res. near equator with gradual transformation to square grid cells away from equator; sea ice : NERSC Sea Ice Model (Viscous- plastic + thermodynamic), 4 gridcells per ocean grid cell; land : ISBA ARPEGE V3 version' history = 'Created from MICOM output using BCM post- processing tools:sigma2z -> vertical interpolation (linear) from isopycnic layer coordinates to standard levels.uvrot -> vector rotation to east, north directions.gen2reg -> horizontal interpolation (linear, gaussian weights) from original curvlinear grid to regular longitude/ latitude grid.' production = 'Created from the ARPEGE postprosessing output files using fortran libnc' institution = 'BCCR (Bjerknes Centre for Climate Research) University of Bergen, Norway (www.bjerknes.uib.no) NERSC (Nansen Environmental and Remote Sensing Center, Norway (www.nersc.no)' Conventions = 'CF-1.0' date = '17-Mar-2005' 152

references = 'www.bcm.uib.no' comment = 'Experiment was initiated from model year 85 (nominal year 1850) of the control simulation (with bcm_experiment_name: PC3). The time varying forcing agents were varied based upon observations and reconstructions for the late 19th and 20th centuries. Varying forcing agents: CO2,CH4,N2O, CFC11(including other CFCs and HFCs),CFC12 and sulfate aerosols (Boucher data, only direct effect). Non-varying forcing agents: Background aerosols: black carbon,sea salt,desert dust, stratospheric and tropospheric ozone, solar irradiance (1368 W/m2) and the distribution of land cover types. No volcanic aerosols. No heat or fresh water adjustments.' project_id = 'IPCC Fourth Assessment' experiment_id = '720 ppm stabilization experiment (SRES A1B)' realization = 1 bcm_experiment_name = 'A1B' contact = 'Email: [email protected], WWW: www.bcm.uib.no' acknowledgment = 'Use of BCM data should acknowledge the contribution of the BCM project and BCM sponsor agencies with the following citation: 'This research uses data provided by the Bergen Climate Model (BCM) project (www.bcm.uib.no) at the Bjerknes Centre for Climate Research, largely funded by the Research Council of Norway.'. The words 'Bergen Climate Model' and 'BCM' should be included as metadata for webpages referencing work using BCM data or as keywords provided to journal or book publishers of your manuscripts.'

Dimensions: lat = 64 lon = 128 bounds = 2 time = 1200 (UNLIMITED) Variables: land Size: 128x64 Dimensions: lon,lat Datatype: int32 Attributes: long_name = 'land mask' lat Size: 64x1 Dimensions: lat

153

Datatype: double Attributes: long_name = 'latitude' units = 'degrees_north' standard_name = 'latitude' bounds = 'lat_bnds' axis = 'Y' lat_bnds Size: 2x64 Dimensions: bounds,lat Datatype: double lon Size: 128x1 Dimensions: lon Datatype: double Attributes: long_name = 'longitude' units = 'degrees_east' standard_name = 'longitude' bounds = 'lon_bnds' axis = 'X' lon_bnds Size: 2x128 Dimensions: bounds,lon Datatype: double tas Size: 128x64x1200 Dimensions: lon,lat,time Datatype: single Attributes: long_name = 'Temperature 2m' standard_name = 'air_temperature' units = 'K' time Size: 1200x1 Dimensions: time

154

Datatype: double Attributes: long_name = 'Time' units = 'days since 1800-01-01 00:0.0' calendar = 'gregorian' delta_t = '0000-01-00 00:00:00' avg_period = '0000-01-00 00:00:00' prev_avg_period = '0000-00-01 00:00:00' standard_name = 'time' axis = 'T' bounds = 'time_bnds' time_bnds Size: 2x1200 Dimensions: bounds,time Datatype: double

155

Appendix A2. Developed Script in MATLAB for exporting 'netCDF' files that contain the future simulations from the ENSEMBLES Project

156

Paper 1

Paparrizos, S., Maris, F., Matzarakis, A., 2016. Sensitivity analysis and comparison of various potential evapotranspiration formulae for selected Greek areas with different climate conditions. Theoretical and Applied Climatology, DOI 10.1007/s00704-15-

1728-z

ABSTRACT Background: Potential Evapotranspiration (PET) is one of the most critical parameters in the research on agro-ecological systems. In cases where the necessary equipment for the measurement of ET is not available, ET can be estimate by theoretical or empirical equations that require simple or analytical data. The computational methods for the estimation of PET vary in data demands from very simple (empirically based), requiring only information based on air temperatures, to complex ones (more physically based) that require data on radiation, relative humidity, wind speed, etc.. Methods: The current research is focused on 3 study areas in Greece that face different climatic conditions due to their location. 12 PET formulae were used, analysed and inter- compared in terms of their sensitivity regarding their input coefficients for the Ardas River basin in north-eastern Greece, Sperchios River basin in Central Greece and Geropotamos

River basin in South Greece.

Results: The results indicated that for the areas that face Mediterranean climatic conditions, the most appropriate method for the estimation of PET was the temperature-based, Hamon's

2nd version (PETHam2). Furthermore, the PETHam2 was able to estimate PET almost similarly to the average results of the 12 equations. For the Ardas River basin the results indicated that both PETHam2 and PETHam1 can be used to estimate PET satisfactorily. The temperature-based equations have proven to produce better results, followed by the radiation-based equations.

Finally, PETASCE which is the most common used PET equation can also be applied occasionall y in order to provide satisfactory results. Conclusions: The literature regarding the sensitivity analysis of various PET formulae is insufficient and this is a fact that highlights the importance of the current study. Water managers who are responsible for planning and adjudicating the distribution of water resources can obtain a thorough understanding of the evapotranspiration process and knowledge about the appropriate method to be used every time to estimate the rates of evapotranspiration. Keywords: Potential Evapotranspiration; empirical PET method; Sensitivity analysis;

Meteorological variables; Greece

157

Paper 2

Paparrizos, S., Maris, F., Matzarakis, A., 2016. Integrated analysis of present and future responses of precipitation over selected Greek areas with different climate conditions. Atmospheric Research, 169: 199-208.

ABSTRACT Background: Assessment of future precipitation variations prevailing in an area is essential for the research regarding climate and climate change. The current paper focuses on 3 selected areas in Greece that present different climatic characteristics due to their location and aims to assess and compare the future variation of annual and seasonal precipitation. Methods: Future precipitation data from the ENSEMBLES anthropogenic climate-change (ACC) global simulations and the Climate Local Model (CLM) were obtained and analysed. The climate simulations were performed for the future periods 2021-2050 and 2071-2100 under the A1B and B1 scenarios. Mann-Kendall test was applied to investigate possible trends. Spatial distribution of precipitation was performed using a combination of dynamic and statistical downscaling technique and Kriging method within ArcGIS 10.2.1.

Results: The results indicated that for both scenarios, reference periods and study areas, precipitation is expected to be critically decreased. Additionally, Mann-Kendall test application showed a strong downward trend for every study area. Furthermore, the decrease in precipitation for the Ardas River basin characterised by the continental climate will be tempered, while in the Sperchios River basin it will be smoother due to the influence of some minor climatic variations in the basins' springs in the highlands where milder conditions occur. Precipitation decrease in the Geropotamos River basin which is characterized by Mediterranean climate will be more vigorous. B1 scenario is appeared more optimistic for the Ardas and Sperchios River basins, while in the Geropotamos River basin, both applied scenarios brought similar results, in terms of future precipitation response. Conclusions: Knowledge of the variations of precipitation over agricultural areas that face different climate conditions is very essential, as the results indicated that the precipitation around Mediterranean is expected to be critically decreased. Additionally, complex climatic conditions and rough topography in combination with diverse social, economic and environmental aspects render the Mediterranean region one of the most vulnerable, in terms of climate change. Keywords: Precipitation; IPCC emission scenarios; climate change; downscaling techniques; spatial interpolation; Greece

158

Paper 3

Paparrizos, S., Matzarakis, A., 2016: Assessment of future climate change impacts on the hydrological regime of selected Greek areas with different climate conditions. Hydrology Research, DOI 10.2166/nh.2016.018

ABSTRACT Background: Assessment of future variations of river streamflow is essential for the research regarding climate and climate change. The study is focused on 3 agricultural areas widespread in Greece with different climate conditions and aims to assess the future response of streamflow and its impacts on the hydrological regime, in combination with other fundamental aspects of the hydrological cycle. Methods: ArcSWAT ArcGIS extension was used to simulate the future response of streamflow. Future meteorological data were obtained from various Regional Climate Models (RCMs), and analysed for the periods 2021-2050 and 2071-2100. The autoregressive integrated moving average model (ARIMA) was used for forecasting of the irrigation water withdrawal. Mann-Kendall test was applied to investigate possible trends.

Results: In all the examined areas, streamflow is expected to be reduced. Areas characterized by continental climate will face minor reductions by the mid-century that will become very intense by the end and thus these areas will become more resistant to future changes. Areas characterized by Mediterranean conditions will be very vulnerable in terms of future climate change. Specifically, the strongest reduction is expected in the Sperchios River basin which will lose almost 1/3 of its streamflow, while the Ardas and the Geropotamos River basin will lose more than 1/3. AET will initially face strong decrease during the first half of the century, but by the end of the century it will face minor increase. Conclusions: Areas characterized by Mediterranean climate conditions will be very vulnerable in terms of future climate change. Reduced precipitation is the main reason for decreased streamflow. Nevertheless, the high values of Actual Evapotranspiration by the end of the century due to increases in air temperature will partly equilibrate the water balance. Since the climate in Mediterranean is one of the most vulnerable in terms of climate change, great caution is needed when making conclusions, as the climate uncertainties results from reduced runoff can be one of the major threats in the contemporary water resources management. Keywords: ArcSWAT; climate change; Greece; hydrological modelling; IPCC scenarios; streamflow

159

Paper 4

Paparrizos, S., Matzarakis, A., 2016. Present and future assessment of Growing Degree Days over selected Greek areas with different climate conditions. Meteorology and Atmospheric Physics, DOI 10.1007/s00703-016-0475-8

ABSTRACT

Background: The determination of heat requirements in the first developing phases of plants has been expressed as Growing Degree Days (GDD). The current study focuses on three selected study areas in Greece that are characterised by different climatic conditions due to their location and aims to assess the future variation and spatial distribution of Growing Degree Days (GDD) and how these can affect the main cultivations in the study areas. Methods: Future temperature data were obtained and analysed by the ENSEMBLES project. The analysis was performed for the future periods 2021-2050 & 2071-2100 with the A1B and B1 scenarios. Spatial distribution was performed using a combination of dynamical and statistical downscaling technique through ArcGIS 10.2.1. Results: The results indicated that for all the future periods and scenarios, the GDD are expected to increase for both applied scenarios in every study area. The increase in the Sperchios River basin will be the highest, followed by the Ardas and the Geropotamos River basins. Moreover, the cultivation period will be shifted from April-October to April-September which will have social, economic and environmental benefits. Regarding the spatial interpolation of the GDD, in the upcoming years the GDD units as well as their area will be expanded and the local cultivations will find more favourable conditions in larger areas and increase their productivity. During the sub-period analysis, the Sperchios and the

Geropotamos River basins are presented with critical differentiations between April-October and April-November results.

Conclusions: In the upcoming years the GDD Expansion of the existing cultivation though, will lead to changes in land use but their extensive expansion will face some difficulties due to the complex topography. On the other hand, in the existing cultivations the growing maturity will be achieved in a shorter period of time and this will further reduce the cultivations costs. Additionally, the decrease of the cultivation period will have a direct effect on the efficient energy use as well as the protection, preservation and the sustainable development of the water resources, since irrigation dominates and constitutes the main cause of water scarcity problems in the certain areas. Keywords: Agriculture; Growing Degree Days; climate change; downscaling spatial interpolation; Greece

160

Paper 5

Paparrizos, S., Maris, F., Matzarakis, A., 2016. Integrated analysis and mapping of aridity over Greek areas with different climate conditions. Global NEST Journal, 18(1): 131 -145.

ABSTRACT

Background: Assessment of aridity conditions prevailing in a certain area is essential for the research on climate and climate change. Greece is characterized by a variety of climatic conditions such as drought conditions or flooding phenomena. The current study focuses on three selected areas within Greece that present different climatic characteristics due to their location and aims to analyse and compare the aridity conditions prevailing in these areas. Methods: Aridity conditions were estimated using the Aridity Index (AI). The analysis was performed using the reference period data for annual, as well as seasonal periods. Mann- Kendall test was applied to investigate possible trends. Spatial distribution of aridity conditions was performed using multi-linear regression techniques and Kriging method within ArcGIS 10.2.1. Results: The results indicated that the study areas face humid conditions, mostly due to the existence of high altitudes. Furthermore, the various climatic conditions are responsible for differentiations in seasonal analysis regarding the aridity conditions. The study areas related to the Mediterranean climate resulted more heterogeneous conditions compared with areas affected by the continental climate. Nevertheless, the created aridity spatial maps of trend analysis presented with differentiations, especially in the mountainous areas were an extreme downward trend is appeared. For the southern investigated area in Crete Island characterized purely by Mediterranean climate, the results were more moderate in terms of aridity conditions.

Conclusions: Through the assessment of the aridity conditions, the emerge of drought conditions to appear is possible so the water resources planning and management must follow very careful steps. The results indicated that great caution is needed in making conclusions regarding the climate in the Mediterranean region. The climate of the Mediterranean region is one of the most vulnerable in terms of climate change with diverse social, economic and environmental impacts. The complex climatic conditions prevailing in both Greece and thus the Mediterranean region play a very important role in the creation of drought conditions, flooding phenomena, etc.. Actions need to be taken in order to prevail against the upcoming effects of climate change. Keywords: Aridity; Aridity Index (AI); multi-linear regression; Ordinary Kriging; Greece

161

Paper 6

Paparrizos, S., Maris, F., Weiler, M., Matzarakis, A., 2016. Analysis and mapping of present and future drought conditions over Greek areas with different climate conditions . Theoretical and Applied Climatology, DOI 10.1007/s00704016-1964-x

A BSTRACT Background: Estimation of drought in a certain temporal and spatial scale is essential for the research on climate and climate change. The current study focuses on 3 selected areas in Greece, Ardas River basin in North-eastern Greece, Sperchios River basin in Central Greece and Geropotamos River basin in Crete Island in South Greece that present different climatic characteristics due to their location and aims to analyse and compare the present and future variation of drought conditions prevailing in these areas.

Methods: The Standardized Precipitation Index (SPI) was used to identify and assess the present and future drought conditions. Future precipitation data were derived from a number of

Regional Climatic Models (RCMs) from the ENSEMBLES European Project.

The climate simulations were performed for the future periods 2021-2050 and 2071-2100 under the A1B and B1 scenarios. Mann-Kendall test was applied to investigate possible trends. Spatial distribution of SPI was performed using a combination of dynamic and statistical downscaling technique and Kriging method within ArcGIS 10.2.1. Results: The results indicated that for both scenarios, future periods and study areas, drought conditions are expected to be more severe in the upcoming years. The decrease of the SPI values in the Sperchios River basin is expected to be the strongest, as it is the only study area that will show a negative balance (in SPI values), regarding the drought conditions. For the Ardas and the Geropotamos River basins, a great decrease of the drought conditions will occur during the 2021-2050 period, while for 2071-2100 period the decrease will continue but it will be tempered. Nevertheless, the situation in all study areas according to the SPI classification is characterized as 'Near-normal', in terms of drought conditions. Conclusions: All study areas will face relatively severe drought conditions in the upcoming years. The conditions will be more intense in the lowlands (plain areas) where all the agricultural activity takes place. Elevation and broadly the topography plays a very important role in the creation of different climatic conditions and directly affects the spatial analysis of the drought conditions.

Keywords: Drought analysis; Standardized Precipitation Index (SPI); IPCC emission scenarios; multi-linear regression; spatial interpolation; Greece

162

Paper 7

Paparrizos, S., Maris, F., Matzarakis, A., 2016. A downscaling technique for climatological data in areas with complex topography and limited data. International Journal of Engineering Research and Development, DOI

ABSTRACT Background: The current study describes a technique for downscaling climatological data in areas with limited or no grid data. In cases where grid data are not available and the researcher is called to operate on a regional or the mesoscale and produce detailed and not coarse results, this technique can be a helping hand. Methods: The described technique constitutes a coupling of statistical downscaling through multi-linear regression techniques with the dynamical downscaling through Geographical Information Systems and it can be used in order to spatially interpolate with high resolution various climatological variables starting from a 1x1km grid. During the statistical analysis, special attention is given to the output p-values, to preserve the statistical significance of the examined factors. Results: The application of the described technique was applied to 3 agricultural areas that present different climate conditions and complex topography. The results indicated that the current technique delivered very sufficient results as the adjusted coefficient (R2) is appeared with high values in almost every case. Areas characterized by Mediterranean type of climate with hot summers (Csa) showed the strongest presumption against null hypothesis, while areas characterized by a combination of different Mediterranean climate types (Csa and Csb) used the most coefficients in the multi-linear procedure and produced relatively good results. Areas facing continental climate conditions delivered satisfactorily results, although most of the examined coefficients are presented with medium presumption against null hypothesis. Conclusions: In areas where complex climate and topographic conditions exist, it is rather difficult to adjust and apply a method that is referred to regional case studies into a larger area, because the under examination coefficients present high variability and they are affected by various parameters each time. The described technique is referred and focused on the mesoscale and it constitutes a simple method, which in combination of statistical analysis and spatial interpolation technique through GIS programs can produce reliable maps that can be notified free of charge to farmers, researchers or various stakeholders. Keywords: Downscaling; Climatological Data; Statistical Analysis; Geographic Information Systems; Greece

163