ENDING OF THE 2006 SUMMER DROUGHT IN THE

LINEKE WOELDERS WAGENINGEN UNIVERSITY, JUNE 2007 WAGENINGEN

Ending of the 2006 summer drought in the Netherlands

Lineke Woelders (831102968020) Wageningen University, June 2007 Wageningen

Course name: Soil, Water, Atmosphere: Bachelor Completion 2 Course code: HWM-80806 (6 ects) Supervisor: Dr. ir. H.A.J. van Lanen Hydrology and Quantitative Water Management Group

Acknowledgement I would like to thank the Royal Netherlands Meteorological Institute (KNMI) for their free and web-accessible data. Without those data I would not have been able to carry out this study. Also I would like to thank my supervisor Henny van Lanen for the access to his personal database on drought, and for sharing his personal knowledge on droughts.

Lineke Woelders, Wageningen, June 2007

i ii

Table of contents 1. INTRODUCTION ...... 7 2. METHODS AND DATA...... 9 2.1 Meteorological data...... 9 2.2 Potential evapotranspiration...... 10 2.3 Meteorological overviews by KNMI ...... 12 2.4 Soil moisture and groundwater recharge ...... 12 2.5 Drought characterization ...... 13 2.6 Procedure...... 15 3. DROUGHTS...... 17 3.1 Precipitation ...... 17 3.2 Cumulative potential precipitation deficit ...... 18 3.3 Comparison with KNMI cumulative potential precipitation deficit maps ...... 20 3.4 Soil moisture in summer ...... 21 3.5 Groundwater recharge in winter...... 25 3.6 Discussion...... 28 4. CONCLUSIONS AND RECOMMENDATIONS...... 31 REFERENCES...... 33 APPENDIX ...... 35

iii

List of figures Figure 2.1. Locations and numbers of the 9 selected KNMI meteorological stations used in this study...... 9 Figure 2.3. Definition of drought deficit characteristics...... 14 Figure 2.4. Flow chart of the data processing...... 15 Figure 3.1. Precipitation in de Bilt ...... 17 Figure 3.2. Cumulative potential precipitation deficit for De Bilt from April 1 until November 22, 2006 ...... 19 List of tables Table 2.1. Soil moisture contents (%) of a clayey soil and a sandy soil at different pF values ...... 13 Table 2.2. Soil moisture storage for the clayey and sandy soil...... 14 Table 3.1. Magnitude, date of occurrence and date of ending of the maximum potential precipitation deficit in 2006 at the investigated stations...... 20 Table 3.2. Comparison of calculated values on potential precipitation deficit with KNMI classes (mm)...... 21 Table 3.3. Drought periods and their intensity at a clayey soil in Leeuwarden and , 2006 ...... 23 Table 3.4. Drought periods and their intensity at a sandy soil in Leeuwarden and Amsterdam, 2006...... 25 List of appendices Appendix 1...... 36 Appendix 2...... 41 Appendix 3...... 43 Appendix 4...... 44 Appendix 5...... 46 Appendix 6...... 47 Appendix 7...... 48 Appendix 8...... 49

iv

Summary In this study, the spatial characteristics of the ending of a drought are investigated. The results are illustrated for the ending of the 2006 summer drought in the Netherlands. Meteorological data and time series of 30 years of 9 meteorological stations are used, which are biased to the North and West of the Netherlands. These data and time series are provided by the Royal Netherlands Meteorological Institute (KNMI). The potential evapotranspiration is calculated for each meteorological station by using the Makkink equation. The precipitation and potential evapotranspiration are used to calculate the cumulative potential precipitation deficit. KNMI provides daily maps of the Netherlands on cumulative potential precipitation deficit. These maps for the summer of 2006 are used in this study to compare the calculated results with the provided KNMI maps. The soil moisture content and groundwater recharge are simulated by using the transient soil water balance model NUT_DAY. The investigated soils are a clayey soil and a sandy soil. For drought characterization, thresholds on the investigated variables are defined. For precipitation and groundwater recharge, the threshold level method is applied. For cumulative potential precipitation deficit, a drought is considered to occur once a deficit occurs. On this variable, a threshold on drought duration is arbitrarily set. For soil moisture, the critical soil moisture is used as threshold. All stations experienced a period of drought in the precipitation in the 2006 summer. In August, all stations received enough precipitation to end the precipitation drought. In September, all stations experienced a second, minor drought. The ending of the 2006 summer drought in the cumulative precipitation deficit varied between the different stations. A single rainfall event can delay or advance the ending of a drought considerably. Also the maximum cumulative potential precipitation deficit can be influenced by single rainfall events that occur at the start of the summer season. The comparison on the calculated time series of the cumulative potential precipitation deficit for the 9 stations with the KNMI maps shows that only Leeuwarden substantially deviates. The cumulative potential precipitation deficit for the other stations mostly is in the class provided by KNMI, or in the next class. The summer soil moisture is considered in particular. The soil moisture storage for Leeuwarden (driest station) and Amsterdam (wettest station) are compared. The clayey soil in Leeuwarden experiences the most severe drought with the longest duration. The sandy soil in Amsterdam experiences the less severe drought with the shortest duration. Soil type (readily available soil moisture) and rainfall influence the soil moisture drought, and therefore the development or ending of a drought. Winter groundwater recharge is considered in particular. The winter before the summer of 2006 might have influenced the summer drought of 2006. For some locations, the summer drought of 2006 might influence the summer of 2007. Several conclusions can be drawn from this study. The 9 selected meteorological stations are biased and therefore not completely representative. It can be concluded that the development or ending of a drought depends on the chosen threshold. The ending of the drought in the cumulative potential precipitation deficit varies from station to station because of single (convective) rainfall events from early August onwards that vary spatially. The development and ending of the soil moisture drought are strongly dependent on soil type. NUT_DAY is not taking surface runoff or soil altering effects due to drought into consideration, which might influence the results.

v

vi

1. Introduction The Netherlands is a country that is well known for its battle with water. Situated in the deltas of the River Rhine, the River and the River Scheldt, the country is flood prone and the population has fought against the sea and river water for thousands of years (Huisman et al., 1998). Today, still, this battle plays an important role in water management in the Netherlands. Despite the fact that a surplus of water seems to be the problem in the Netherlands, the country faces droughts from time to time as well. This rather complex water related problem can cause a lot of problems in the Netherlands, because a drought can have large spatial extent, many actors are involved and future forecasts are unsure. Future precipitation anomalies for 2050 are predicted to be raised by an average of 6%, but by 75% in a dry scenario (RIZA et al., 2005). The reason that drought problems are complex is very well described by Tannehill (1947):

"We have no good definition of drought. We may say truthfully that we scarcely know a drought when we see one. We welcome the first clear day after a rainy spell. Rainless days continue for a time and we are pleased to have a long spell of such fine weather. It keeps on and we are a little worried. A few days more and we are really in trouble. The first rainless day in a spell of fine weather contributes as much to the drought as the last, but no one knows how serious it will be until the last dry day is gone and the rains have come again... we are not sure about it until the crops have withered and died."

So while we can be often quite sure we are actually in a dry period, the question is what really defines the fact that we are in a drought (Tallaksen and van Lanen, 2004). A dry period can sometimes pass by quite unnoticed, for instance, if its duration is short enough. Also its intensity might be low enough so that the dry spell does not cause too much problems. But sometimes, dry periods can have large impacts on agricultural crops and vegetation because the opposite is true: they have a duration that is long enough, they occur in the growing season and/or have a large intensity. The drought that stroke the Netherlands in 2006 is an example of a drought that had a large impact (Lanen, 2006). It lasted almost 2 months, during the growing season. Almost no rain was recorded in that period (KNMI, 2006). After those 2 months, in August, it started raining again. One might take this moment as marker point for the ending of the drought, but if the cumulative potential precipitation deficit (the balance between precipitation and potential evapotranspiration over a period of time) is considered, the drought might not end at the same moment. Also, if the soil moisture and groundwater recharge are considered, the drought might end later. Identifying the ending of a drought is investigated and discussed in this paper, with special attention to the 2006 summer drought in the Netherlands.

Objectives This paper has 2 main objectives:

1. To investigate the spatial characteristics of the ending of a drought, which might depend on duration, intensity, soil type and season; 2. To illustrate the results for the ending of the drought in the Netherlands for the summer of 2006.

Approach To study the ending of a drought, first, we must be sure we are actually dealing with one. For a meteorological drought (Tallaksen and van Lanen, 2004), a threshold on the precipitation 1 and potential precipitation deficit needs to be selected. This threshold can then be compared with time series on precipitation and the precipitation deficit for a specific year with a possible drought, like 2006, to study the ending of the meteorological drought. For the hydrological drought, the simulated soil moisture storage and groundwater recharge are considered. Thresholds on these need to be selected as well, after which they can be compared with time series on soil moisture and groundwater recharge for a specific year with a possible drought, like 2006.

Outline In chapter 2 the data and methods are discussed. Thirty years of daily meteorological measurements from 9 KNMI meteorological stations are considered in this study. These data are used to calculate reference evapotranspiration, the cumulative potential precipitation deficit, and to simulate soil moisture storage and groundwater recharge. Subsequently, thresholds on these variables are set. The results are given in chapter 3. The thresholds on precipitation, cumulative potential precipitation deficit, soil moisture and groundwater recharge are applied to the time series from the drought of 2006 in the Netherlands and discussed. The paper ends with conclusions and recommendations in chapter 4.

8 2. Methods and data

2.1 Meteorological data The meteorological data that are used in this paper are provided by the Royal Netherlands Meteorological Institute (KNMI). The KNMI has in total 325 manned and unmanned meteorological and precipitation stations in the Netherlands, but only 9 manned stations are considered in this study. The reason for that is that not all stations provide data which are easy accessible through the web and useful for this research, or provide time series that are long enough. Time limitations for this study have also played a role in the choice of the number of meteorological stations. The locations and official numbers of the 9 selected meteorological stations can be found in figure 2.1. Unfortunately, the selected stations are biased to the North and West of the Netherlands. Not more than 2 stations are located in the East and Southeast of the Netherlands.

270 280

235

240 290 260 Figure 2.1. Locations and numbers of the 9 selected KNMI 344 meteorological stations used in this study. 235=Den Helder (de Kooy); 240= Amsterdam (Schiphol); 310 260=De Bilt; 270=Leeuwarden; 280= (Eelde); 290=Twenthe; 380 310=Vlissingen; 344=; 380=Maastricht (Beek).

The data from these stations that are used in this research are: • Latitude of the station (rad) • Daily mean temperature (°C) • Daily minimum temperature (°C) • Daily maximum temperature (°C) • Daily mean surface air pressure (hPa) • Sunshine duration (0.1 h) • Daily precipitation amount (0.1 mm)

All these data (except for the latitude) are provided daily for each station. Per station a time series of 30 years of daily data is used, from January 1st, 1977 or until either December 31, 2006 or March 1st, 2007. The series from January 1977 until December 2006 are used to calculate the 30 year averages, whereas the time series after December 2006 are used for the assessment of the possible impact of the 2006 drought on 2007.

9 Although almost all stations provide a continuous time series of data, Leeuwarden, Twenthe and Vlissingen have missing data on sunshine duration. Leeuwarden misses 10 months in 1993, Twenthe misses all months in 1993 and Vlissingen misses 6 months in 1994. To use these three time series, the missing values for sunshine duration are replaced by measurements from a meteorological station nearby. Leeuwarden uses measurements from Den Helder, Twenthe uses measurements from Groningen and Vlissingen uses measurements from Rotterdam. The reason that these stations and no others are chosen is that these stations are geographically closest to the station with missing data. Also, the measurements on sunshine duration three years prior to the missing data are checked. The selected stations resemble the stations with missing data more than other stations.

2.2 Potential evapotranspiration Every month, the KNMI publishes the MOVN. The MOVN is a monthly overview on daily data on precipitation and reference evaporation for a number of meteorological stations (KNMI, 2006). In this study, the data on precipitation that can be found in this overview will be used, as they can also be found on the web. The data on reference evaporation, however, will not, and even can not, be used. The reasons for this are that these data are not found on the web, and therefore should be manually processed which would take too much time for this study. Moreover, the KNMI has published data on reference evapotranspiration for 5 meteorological stations only. This would not be enough for this study. Finally, the 5 time series the KNMI provides on reference evaporation are not long enough for this study. Therefore, the reference evaporation for the 9 meteorological stations needs to be calculated This is done by using the Makkink equation (eq. 2.1) (Makkink, 1957), which is commonly used in the Netherlands. After that, the potential evapotranspiration can be calculated from the reference evapotranspiration. With these data, eventually, the cumulative potential precipitation deficit (precipitation – potential evapotranspiration) can be calculated for each station to investigate possible drought development.

 s    λwETref = 0.65 Rs (eq. 2.1)  s + γ air  -1 in which: λw is latent heat of vaporization (MJ kg ) -2 -1 ETref is reference evapotranspiration (kg m d ) s is the slope of the vapour pressure curve (kPa °C-1) -1 γair is the psychrometric constant (kPa ) -2 -1 Rs is global radiation (MJ m d )

These unknown variables can be calculated as follows (Dam et al., 2005):

−3 λw = 2.501− 2.361∗10 Tair (eq. 2.2) in which Tair is the mean daily air temperature.

4098∗ esat s = 2 (eq. 2.3) ()Tair + 237.3 in which esat is saturation vapour pressure (kPa).

pair γ air = 0.00163∗ (eq. 2.4) λw in which pair is surface air pressure (in kPa).

10 Rs = ()a + bN rel Ra (eq. 2.5) in which a and b are parameters between 0 and 1, Nrel is the relative sunshine -2 -1 duration (-) and Ra is extraterrestrial radiation (MJ m d ).

 17.27∗Tair ,min 17.27∗Tair ,max  e = 0.305e Tair ,min +237.3 + e Tair ,max +237.3  sat   (eq. 2.6)   in which Tair,min and Tair,max are the minimum and maximum daily air temperature.

Nact Nrel = (eq. 2.7) Nmax in which Nact is the actual sunshine duration, and Nmax is the maximum sunshine duration.

G R = sc d []ω sin(ϕ)sin(δ ) + cos(ϕ)cos(δ )sin(ω ) (eq. 2.8) a π r s s in which: Gsc is the solar constant (118.1 MJ m-2d-1) dr is inverse relative distance Earth-Sun (-) ωs is sunset hour angle (rad) φ is latitude (rad) δ is solar declination (rad)

 24  N max =  ωs (eq. 2.9)  π 

 2π  dr = 1+ 0.033cos J  (eq. 2.10)  365  in which J is Julian day.

 2π  δ = 0.409sin J −1.39 (eq. 2.11)  365 

ωs = arccos[]− tan(ϕ)tan(δ ) (eq. 2.12)

The Makkink equation (eq. 2.1) computes the ETref. Multiplying this reference evapotranspiration with a crop factor gives the potential evaporation ETpot for a certain crop. In this research, because of time limitations, only grass is investigated. The crop factor of grass is approximately 1 all year round (Feddes, 1987), so ETpot = ETref. It should be noted that KNMI calculates sunshine duration for the ETref in the MONV from global radiation with an algorithm (Slob and Monna, 1991). Because KNMI only provides data on sunshine duration on their website, in this study, global radiation is calculated from the measured sunshine duration by KNMI. This is done by using eq. 2.5. For global use, Allen et al. (1998) recommend values of respectively 0.25 and 0.50 for the parameters a and b in eq. 2.5.

11 2.3 Meteorological overviews by KNMI Besides the meteorological data and time series, which can be found on the website from KNMI, also some maps on the cumulative precipitation deficit from KNMI of the meteorological situation for 2006 are examined. The calculated time series of on potential precipitation deficit are compared with these maps, to check the whether the results are representative and to understand the outcomes better. Although these maps were provided by KNMI on their website, they are no longer available on the website as KNMI refreshes these daily maps every day. Two key maps on the cumulative potential precipitation deficit in 2006 are given below (figure 2.2a and 2.2b). The cumulative potential precipitation deficit runs from April 1st, 2006 until July 30 and August 26, respectively. The 1st of April is the start of the hydrological summer season and is assumed to be the start of the growing season. The deficit is presented in different classes with contrasting colours. More maps on the cumulative potential precipitation deficit can be found in Appendix 1. Only 11 maps will be examined in this study. These are the maps for July 20, 25 and 30, for August 5, 10, 15, 21, 26 and for September 4, 14 and 28.

(a) (b) Figures 2.2a and 2.2b. Potential cumulative precipitation deficits from the beginning of the h ydrological year 2006 until a: July 30 and b: August 26. 2.4 Soil moisture and groundwater recharge To model soil moisture and groundwater recharge, a model called NUT_DAY (Peters et al., 2001) is used. This model does not solve the soil water flow equation but only the water balance (Lanen et al., 1996). NUT_DAY has been developed for daily time steps. In NUT_DAY, the daily balance of the root zone is written as follows:

r r a r St = St−1 + Pt − ETt − Qt (eq. 2.13) r in which S t = soil moisture storage in the root zone at the end of day t (m) r S t-1 = soil moisture storage in the root zone at the end of day t-1 (m) Pt = precipitation day t (m) a ETt = actual evapotranspiration day t (m) r Q t = precipitation excess day t (m)

12 As can be seen, the model does not include capillary rise. This is because the model is developed for situations with deep groundwater tables. For each meteorological station, daily, monthly and yearly data on soil moisture storage and groundwater recharge are simulated using this model. The input for this model are time series of daily precipitation and on potential evaporation, and soil data. The precipitation data are the daily precipitation measurements obtained directly from the meteorological stations. The daily potential evapotranspiration time series are calculated by the Makkink equation (chapter 2.2). The soil data needed by NUT_DAY can be divided in general data and specific data. The general data consist of soil thickness, number of soil layers and rooting depth. The soil is assumed to be 1 m thick, and to consist of only one layer. Furthermore, it is assumed that the only crop is grass, and that the rooting depth is 50 cm. The specific soil data that are needed by the model are soil moisture percentages at pF 2 (field capacity), pF 3 (critical point) and pF 4.2 (wilting point). In this paper, two soil types are investigated which can be found in table 2.1, light clay that is very low in organic matter and moderate to very fine, slightly loamy sand that is extremely low in organic matter.

Table 2.1. Soil moisture contents (%) of a clayey soil and a sandy soil at different pF values, derived from Locher and de Bakker (1990) pF Soil type

Light clay, very low in organic Moderate to very fine, slightly loamy matter sand, extremely low in organic matter 2 41 19 3 34 7 4.2 25 3

2.5 Drought characterization Threshold level method As explained by Hisdal et al. (2004), the threshold level method can be used to define droughts. This method is the most frequently applied quantitative method where it is essential to detect the beginning and the end of a drought. In the paper by Hisdal et al. (2004) the threshold is based on a certain discharge Q below which river flow is considered as a drought. Figure 2.3 shows a threshold level Q0 which is fixed for the period of record. In this study, the used threshold level is a monthly varying threshold. Events defined with the varying threshold should be called anomalies rather than droughts. In this study, however, also the values below the monthly varying threshold are considered as a drought. The threshold level method is applied to precipitation, the cumulative potential precipitation deficit and groundwater recharge. If values on these are below a certain threshold, a drought can occur. According to Hisdal et al. (2004), for Q, a range of thresholds from Q70 to Q95 is considered reasonable for perennial streams. The threshold that is arbitrarily chosen in this paper is X90 (30-day), which means that precipitation and groundwater recharge must be lower than 90% of the 30 year monthly sums to be considered a drought.

13

)

-1 s 3

Flow (m

d 1 d 2 d 3 d 4 Q v v 0 v 1 v 2 3 4 Q min Time (days) Figure 2.3. Definition of drought deficit characteristics (Hisdal et al., 2004).

Cumulative potential precipitation deficit Once a cumulative potential precipitation deficit occurs instead of a surplus, a drought is considered to occur. Here, a threshold on duration of drought is arbitrarily chosen; i.e. 7 days. This means that, for instance, a cumulative potential precipitation deficit that lasts less than 7 days will not be considered a drought.

Soil moisture storage Below the critical point of soil moisture, vegetation can not evapotranspire at the potential level. In this study, when the soil moisture storage drops below the critical value (table 2.1), a drought is considered to begin. The sensitivity of a soil for this to happen is dependent on the readily available soil moisture storage, which is the difference between the soil moisture storage at the critical point and the soil moisture storage at field capacity. The readily available soil moisture storage should not be confused with the available soil moisture storage. The latter is the difference between the soil moisture storage at wilting point and the soil moisture storage at field capacity. For the two investigated soils, the soil moisture storages in mm are presented in table 2.2.

Table 2.2. Soil moisture storage for the clayey and sandy soil Soil type

Variable Light clay, very low in organic Moderate to very fine, slightly matter loamy sand, extremely low in organic matter Soil moisture storage at field 205 95 capacity (mm) Critical soil moisture storage 170 35 (mm) Readily available soil 35 60 moisture (mm) Soil moisture storage at 125 15 wilting point (mm) Available soil moisture 80 80 storage (mm)

14 2.6 Procedure Figure 2.3 presents the flow chart of the investigation process. Different paths are followed for calculating the different variables that are investigated as explained in the chapters 2.1 to 2.5. For monthly precipitation, the path is as follows: 3 → 12 →16 For the daily cumulative potential precipitation deficit, the path is as follows: 4 + (5 → 8) → 14 For soil moisture, two paths are followed. The first path is the calculation of the 30 year monthly averages and the monthly averages of 2006: 1 + 2 + (5 → 8 → 7) → 10 → 15 →11 →13 The second path is the calculation of daily values in 2006: 1 +2 + (5 → 8 → 7) → 10 → 15 → 17 For monthly groundwater recharge, the path is as follows: 1 +2 + (5 → 8 → 7) → 6 → 9 → 12 → 16

General and specific Daily P Other meteorological soil data variables (except for P)

1 2 3 4 5

NUT_DAY Daily ETref

6 7 8 Daily Aggregate monthly Daily P-PET groundwater 9 recharge

10 11 12 13 14

Daily soil moisture Calculate X90

15 16 Critical soil Drought evaluation moisture values 17

Figure 2.4. Flow chart of the data processing

15 16 3. Droughts

3.1 Precipitation

The X90 of the 30 year monthly precipitation sums for the 9 selected meteorological stations and the monthly averages per station for 2006 are compared. A monthly average in 2006 below the X90 of the 30 year monthly precipitation sum is considered a drought. The reason why monthly and not daily values are considered, is that precipitation is very irregular on a daily basis. For each station, graphs are made which show the 30 year monthly average, the X90 of the 30 year monthly precipitation sum and the monthly average for 2006 (figure 2.4). An example of such a graph is figure 3.1 for the de Bilt station. The graphs of the other stations are found in Appendix 2.

Figure 3.1. Precipitation in de Bilt

As can be seen, the 30 year monthly averages on precipitation for the 9 stations are very much alike, somewhere between 50 and 100 mm a month. Also, the X90 of the 30 year monthly precipitation sum shows at each station a somewhat similar pattern, i.e. between 0 and 50 mm a month. For the beginning of the calendar year 2006, the amounts of precipitation that each station received also show a similar pattern. Each station is more or less dry in January, and catches above average precipitation in May. In June, most stations (except for den Helder) are quite dry. Twenthe, Rotterdam, Groningen and de Bilt are dry enough to be considered a drought during that period. In July, most stations, except for Twenthe and Maastricht (which is only 0.6 mm wetter than X90), are drier than X90 and droughts are considered to occur at these stations. In August, precipitation for the stations varies considerably, from 80 mm in Leeuwarden to 268 mm in Rotterdam. A precipitation pattern seems to occur. The stations located in the western part of the Netherlands, i.e. Vlissingen, Rotterdam, Amsterdam and den Helder, receive relatively high precipitation in August (in a range from 268 mm in Rotterdam to 201 mm in den Helder). The stations that are situated more inland, like de Bilt, Groningen, Twenthe and Maastricht, receive lower precipitation (in a range from 167 mm in Twenthe to 181 mm in de Bilt and Maastricht). Leeuwarden misses a peak in precipitation in August that all the other stations do show. All 9 stations received above the 30 year monthly average precipitation. According to precipitation, the summer drought of 2006 ends in August, even in Leeuwarden.

17 In September, all stations show a precipitation that is below X90 and therefore a second, minor drought in precipitation occurs. October is for every station wet or normal (above the 30 year average or below the 30 year average but above X90), as well as November and December.

3.2 Cumulative potential precipitation deficit Potential evapotranspiration To calculate the cumulative potential precipitation deficit, the potential evapotranspiration needs to be calculated first. The potential evapotranspiration is derived from the reference evapotranspiration ETref (chapter 2.2). For the 9 meteorological stations, the ETref was calculated by using the Makkink equation (eq. 2.1). Global radiation was calculated from sunshine duration by eq. 2.5, and Allen et al. (1998) recommends 0.25 and 0.50 for a and b, respectively, for global use. To check if this assumption is correct, the calculated reference evaporation and the official KNMI values for the months May, June and July of 2006 (KNMI, 2006) for station de Bilt (which is one of the 5 stations for which KNMI provides the reference evaporation in the MONV) was compared. The calculated reference evaporation considerably deviates with the official values given by KNMI in the MONV. This deviation probably is caused by the use of eq. 2.5 to calculate global radiation instead of the algorithm proposed by Slob and Monna (1991). The deviation could be minimized, however, by optimizing parameters a and b in eq. 2.5. This is done by using the sum of squares method (SS). The SS is the sum of the squares of the differences of the predicted values and the grand mean:

n ˆ 2 ∑ ()yi − y (eq. 3.1) i=1 in this study: ŷi = the calculated reference evaporation y = the reference evaporation given by KNMI

The difference between ŷi and y , and thus the SS, is minimized by optimizing parameters a and b in eq. 2.5 by using Solver in Excel. With these optimized parameters, the calculated reference evaporation resembles the official KNMI values better. A graph showing the official KNMI values, the calculated values according to Allen et al. (1998) and the calculated values with the optimized parameters can be found in Appendix 3. The optimized values for the De Bilt station are then implemented in the equations for the other meteorological stations to improve reference evaporation for every station. An example of this improved outcome is found in Appendix 3, where the optimized values derived from de Bilt are implemented in the equation for den Helder. It is assumed that the parameters found for de Bilt are also the best choice for implementation in the other stations because of the central geographical position of de Bilt. As can be seen for den Helder, the cumulative difference between the official KNMI values and the calculated values according to Allen et al. (1998) is 127 mm. It should be noted that although the optimized parameters provide a better result, the cumulative difference between the calculated reference evaporation and the official KNMI data is still 66 mm.

Cumulative potential precipitation deficit For the 9 meteorological stations, the cumulative potential precipitation deficit (cumulative precipitation minus cumulative potential evapotranspiration) starting at the beginning of the hydrological year (April 1st) was calculated for 2006. Because the deficit is calculated cumulatively, daily values can be used instead of monthly values. The results are plotted for

18 each meteorological station (figure 3.2). The cumulative potential precipitation deficit starts at April 1st and ends at the last day of the precipitation deficit, which is November 22 for De Bilt (figure 3.2). The graphs for the other 8 stations can be found in Appendix 4.

Figure 3.2. Cumulative potential precipitation deficit for De Bilt from April 1 until November 22, 2006

The graphs of the cumulative potential precipitation deficit (called ‘precipitation deficit’ in the following) for 2006 show a pattern that seems to be quite the same for each station, neglecting some incidental precipitation events. In general, a precipitation deficit occurs from the beginning of the hydrological year until half of May. Then, a period with precipitation happens that ends at the beginning of June. This period with precipitation is generally not sufficient to result in a cumulative precipitation surplus. After that, a very dry period occurs that lasts until the beginning of August. At that moment, it starts raining again until the beginning of September. The whole September is dry again and in the beginning of October it starts raining. Although this pattern can generally be found in the graph of the precipitation deficit for every station, there are some differences that cause the large differences in the end of the drought in the precipitation deficit. These differences are mostly due to the large differences at every station in the precipitation event that took place at the beginning of August. Moreover, the precipitation sum that the stations have received in the last half of May plays a role. This precipitation sum in the second part of May differs from station to station, causing a small difference in the maximum precipitation deficit in summer. Maastricht, for example, receives so much precipitation in the second part of May a cumulative potential precipitation surplus happens at the end of May. The maximum precipitation deficit in Maastricht is 165 mm, which is the lowest of all stations. The graphs (Appendix 4) at the end of August show that the maximum precipitation deficit does not differ so much from station to station ( table 3.1), except Maastricht. The coastal region seems to be slightly drier. During the rainfall in August, a more clear pattern becomes visible because in general the inland stations receive systematically less precipitation than the stations near the coast (except Vlissingen). This results in precipitation deficits inland that clear up later than the precipitation deficits near the coast ( table 3.1).

19 Table 3.1. Magnitude, date of occurrence and date of ending of the maximum potential precipitation deficit in 2006 at the investigated stations Meteorological station Maximum potential Date of maximum Date of ending of precipitation deficit potential precipitation potential precipitation (mm) deficit (2006) deficit (2006) Den Helder 215 July 31 November 17 Amsterdam 191 July 31 Augustus 28 Rotterdam 239 July 31 October 21 Vlissingen 229 July 31 December 11 De Bilt 216 July 29 November 22 Leeuwarden 250 October 1 January 16, 2007 Groningen 209 July 29 December 29 Twenthe 201 July 31 December 10 Maastricht 163 July 29 December 6

But there are also differences between stations in a specific region (inland or coast). Leeuwarden, for example, misses the large precipitation in August that has been recorded in all other stations (chapter 3.1). This leads to a much larger precipitation deficit that occurs later that year. The maximum precipitation deficit in Leeuwarden occurs at the end of September and is 250 mm, whereas the other stations have their maximum deficit, mostly around 200 mm, already at the end of July. The deficit at Leeuwarden disappears much later than the deficits for the other stations. The station Leeuwarden receives from the first of October on comparable precipitation as the other stations do. The potential precipitation deficit of Amsterdam, on the other hand, disappears relatively fast because of very high precipitation in August. It should be noticed, though, that in September a new, small precipitation deficit occurs that lasts until early October.

3.3 Comparison with KNMI cumulative potential precipitation deficit maps Every day, KNMI publishes on the website a map for the Netherlands on which the cumulative potential precipitation deficit for the previous day is given. On those maps, the Netherlands is presented in certain classes. The KNMI maps of 11 days in 2006 are compared visually with the calculated precipitation deficit for these days. The results are found in table 3.2.

20 Table 3.2. Comparison of calculated values on potential precipitation deficit with KNMI classes (mm). Difference with KNMI = the amount of mm the calculated value differs from the KNMI class on the map. No value: calculated value lies within the class provided by the KNMI. Light blue value: calculated deficit is x mm smaller than the given KNMI class, and lies in the previous class provided by KNMI Light red value: calculated deficit is x mm larger than KNMI class, and lies in the next class provided by KNMI Dark red value: calculated deficit is x mm larger than KNMI class, and lies two classes or more above the class provided by KNMI July July July Aug. Aug. Aug. Aug. Aug. Sept. Sept. Sept. 20 25 30 5 10 15 21 26 4 14 28 Calculated values -178 -196 -212 -175 -172 -125 -127 -111 -92 -118 -139 Den Helder Difference with KNMI 28 19 Calculated values -160 -196 -188 -159 -164 -76 -44 -20 6 -21 -37 Amsterdam Difference with KNMI Calculated values -203 -224 -235 -191 -197 -106 -70 -52 -34 -61 -76 Rotterdam: Difference with KNMI 3 24 2 Calculated values -194 -212 -226 -169 -179 -151 -130 -89 -90 -150 -87 Vlissingen: Difference with KNMI 30 11 Calculated values -186 -205 -211 -180 -183 -144 -113 -121 -98 -127 -147 De Bilt Difference with KNMI 21 Calculated values -182 -199 -210 -208 -208 -205 -210 -216 -203 -226 -247 Leeuwarden Difference with KNMI 8 45 50 66 83 76 127 Calculated values -187 -197 -203 -204 -199 -177 -143 -119 -93 -115 -135 Groningen: Difference with KNMI Calculated values -176 -190 -199 -166 -169 -136 -98 -107 -99 -123 -145 Twenthe Difference with KNMI 16 9

Calculated values -139 -156 -160 -138 -141 -89 -65 -63 -44 -73 -87 Maastricht Difference with KNMI 4 15 33

For Amsterdam and Groningen, all calculated precipitation deficits are in the same class that KNMI provides. All other stations have one or more calculated precipitation deficits that are outside the class KNMI provides. Mostly, these values are only a few mm outside the given class, but some calculated values deviate considerably with the given class. Leeuwarden has the station with the most deviating values. Probably, most deviations are due to the calculation of the global radiation (chapter 3.2). Furthermore, in this study, data from only 9 meteorological stations are used for the calculations, whereas KNMI uses data from about 200 to 250 precipitation stations to calculate the potential precipitation deficit (Appendix 1). Clearly, this can lead to differences in the results. Furthermore, it should be noted that the selected stations are biased (chapter 2.1) and therefore not representative for the Netherlands in general. The KNMI maps show more classes and classes with larger precipitation deficits than can be found in the table. This leads to an underestimation of the precipitation deficit drought according to the calculations.

3.4 Soil moisture in summer During winter time, evapotranspiration is much smaller than during summer time. This causes soil moisture to be at field capacity during most of the winter, but soil moisture contents will drop in summer. Therefore, by investigating droughts, the soil moisture balance of the summer is most relevant.

21 Because of time limitations, soil moisture storage is simulated for two stations. The selected stations that are chosen are Leeuwarden and Amsterdam, because their 2006 summer precipitation deficits disappear the fastest (August 28, 2006) and the slowest (January 16, 2007) respectively (Appendix 4). Soil moisture storages are simulated on a daily basis and aggregated to monthly values (figure 2.4). The monthly graphs are better comparable with other monthly graphs, for example on precipitation. Moreover, monthly graphs give a good first indication of the drought condition. The daily graphs, then, are used to zoom in at remarkable features that are found in the monthly graphs. In this study, understanding the mechanism behind soil moisture is more important than to give a soil moisture storage as close to the real storage as possible. That is why, although the reliability of the evapotranspiration for Leeuwarden might not be as high as for the other stations (table 3.2), this station is still selected because of its more extreme values which are academically more appealing. Furthermore, the simulation with NUT_DAY itself already is based on so many assumptions (chapter 2.5) that it will never give a highly reliable soil moisture storage. For both stations, and the two soils, the 30 year monthly average and the monthly average for 2006 of soil moisture storage is simulated by NUT_DAY and aggregated afterwards (figure 2.4). Moreover, the daily soil moisture storage is simulated daily for both stations and the two soils, for the period April 1 until December 31, 2006. The simulated soil moisture storages can be found in Appendix 5.

Clayey soil – a general observation The available soil moisture storage for the investigated clayey soil is much higher than for the investigated sandy soil (e.g. 205 versus 95 mm at field capacity). However, the amount of readily available soil moisture is lower for the clayey soil (e.g. 35 versus 60 mm). Figure 3.3 and 3.4 provide the 30 year average monthly soil moisture storage and the monthly storage for 2006. The storage is presented as a relative value, implying that the soil moisture value at critical point is set at zero. It becomes clear that even in a normal year (taking the 30 year average as normal), the investigated clayey soil would experience soil moisture stress during the months May, June, July and August. The simulated soil moisture storage is below 0 during these months, and therefore grass cannot evapotranspire optimally. Soil moisture stress occurs at a soil moisture storage beneath the critical soil moisture. The simulated daily soil moisture storages give a more detailed picture of what is happening in 2006 (Appendix 6). Both stations do not experience a relative soil moisture storage for the clayey soils lower than -45 mm. This reflects the highest soil moisture stress possible. This is clear because the soil is at wilting point (table 2.2). The graphs for the clayey soil seem to flatten out as they reach wilting point.

Figure 3.3. Soil moisture storage (mm) for clayey soil in Leeuwarden, from January until December, 2006. (y=0 represents the soil moisture storage at the critical point)

22

Figure 3.4. Soil moisture storage (mm) for clayey soil in Amsterdam, from January until December, 2006. (y=0 represents the soil moisture storage at the critical point)

Clayey soil - Leeuwarden and Amsterdam At the station in Leeuwarden, the clayey soil experiences soil moisture stress in May 2006, but has still a higher soil moisture storage than the 30 year average. The months June to October are drier than the 30 year average. During these months, soil moisture stress occurs, which does not happen in Spetember and October under normal conditions. In November and December the soil moisture storage is at average level. At the station in Amsterdam, in May 2006, the relative soil moisture storage is just above 0 mm and therefore Amsterdam is not experiencing soil moisture stress that month. In June, the relative soil moisture storage in 2006 is higher than during an average year, but below 0 mm and therefore experiences soil moisture stress. In July, the relative soil moisture storage is lower than the average soil moisture content, and they are both below 0 mm. In August, the relative soil moisture storage is much higher than during an average year and well above 0 mm, so the period with soil moisture stress has ended. From August onwards, soil moisture storages are about average. To get a more detailed picture for the 2006 summer, the simulated daily soil moisture storages in 2006 of the clayey soil in both Leeuwarden and Amsterdam are studied (Appendix 6). In Leeuwarden, 2 drought periods can be distinguished and 3 drought periods in Amsterdam. Note that a period of less than the threshold of 7 days (chapter 2.5) with a relative soil moisture storage above 0 mm will be pooled with the adjacent droughts to be classified as one long drought. On the contrary, a dry period less than 7 days is not classified as a drought. The duration of the droughts and their intensity are shown in table 3.3. It is clear that less precipitation causes droughts of longer durations and larger soil moisture stresses.

Table 3.3. Drought periods and their intensity at a clayey soil in Leeuwarden and Amsterdam, 2006 Leeuwarden Amsterdam Period Max. soil moisture Period Max. soil moisture stress (mm) stress (mm) May 8 – May 20 - 21 May 5 – May 20 - 25 June 2 – October 21 - 40 June 9 – August 10 - 37 September 17 – October 1 - 9

Sandy soil – a general observation The maximum amount of readily available soil moisture in the sandy soil is higher than in the clayey soil (table 2.2). During a normal year (taking the 30 year average as normal), the sandy soil supplies sufficient soil moisture every month of the year for both stations, according to calculations with NUT_DAY (figures 3.5 and 3.6).

23 The graphs of the daily relative soil moisture storages (Appendix 6) for both stations on the sandy soil show that soil moisture stress gets no lower than 20 mm. This is obvious, because at this point, no available soil moisture is left (table 2.2). The curves seem to flatten out as they reach the soil moisture storage at wilting point.

Figure 3.5. Soil moisture storage (mm) for sandy soil in Leeuwarden, from January until December, 2006. (y=0 represents the soil moisture storage at the critical point)

Figure 3.6. Soil moisture storage (mm) for sandy soil in Amsterdam, from January until December, 2006. (y=0 represents the soil moisture storage at the critical point)

Sandy soil – Leeuwarden and Amsterdam At the station in Leeuwarden, in 2006 (figure 3.5), the sandy soil experiences soil moisture stress from June until October. In November and December soil moisture storages are around average again. At the station in Amsterdam (figure 3.6), the simulated situation is different. Soil moisture storage is high enough until June. Only July experiences soil moisture stress, followed by a wet August with soil moisture storages much higher than normal. From September onwards, soil moistures are about normal. The graphs of the simulated daily relative soil moisture storage for both Leeuwarden and Amsterdam (Appendix 6) show a more detailed picture of what is happening during 2006. In Leeuwarden, 2 droughts can be distinguished and 1 in Amsterdam. Minor droughts of less than 7 days are pooled. The duration and intensity of these droughts are found in table 3.4. It is clear that less precipitation causes droughts of longer duration and larger soil moisture stress.

24 Table 3.4. Drought periods and their intensity at a sandy soil in Leeuwarden and Amsterdam, 2006 Station

Leeuwarden Amsterdam Period Max. soil moisture Period Max. soil moisture stress (mm) stress (mm) June 7 – August 27 - 19 June 12 – August 1 -18 September 8 – October 4 - 16

Comparison of the clayey and sandy soils The investigated clayey soil has a higher total soil moisture storage than the investigated sandy soil. However, as readily available soil moisture is higher in the sandy soil, this soil is less sensitive to dry periods than the clay soil. The lowest possible soil moisture stress shows this; the soil moisture stress of the clayey soil gets no lower than 45 mm, whereas the stress of sandy soil gets no lower than 20 mm. Consequently, fewer drought periods, also with a lower intensity, are experienced on the sandy soil. The duration of a drought period is shorter as well. For both soils, high precipitation, like the one that occurred in August 2006 in Amsterdam, can instantly increase soil moisture storages and end a drought immediately.

3.5 Groundwater recharge in winter If there is not enough groundwater recharge in the winter season, then droughts might extend in the following summer. Therefore, in this study, the focus is on winter groundwater recharge. Particularly the groundwater recharges of the winter before and after the summer of 2006 are considered. A monthly groundwater recharge that is lower than 90% of the 30 year monthly groundwater recharges is considered as a drought. For the stations Leeuwarden and Amsterdam, that are selected because they are the driest and the wettest station, respectively (chapter 3.3), groundwater recharge for both the clayey soil and the sandy soil are simulated by NUT_DAY. The 30 year monthly average is then calculated, as well as the X90 and the monthly average for 2006. Although the simulation model uses a daily time step, a monthly time step is chosen for the drought analysis. This is done because the daily values of groundwater recharge show a very irregular pattern due to the irregular pattern of precipitation.

Clayey soil The simulated groundwater recharges in a clay soil for both stations Leeuwarden and Amsterdam are presented in figures 3.7 and 3.8.

Figure 3.7. Groundwater recharge (mm) for clayey soil in Leeuwarden, from January until December, 2006

25

Figure 3.8. Groundwater recharge (mm) for clayey soil in Amsterdam, from January until December, 2006

The 30 year monthly averages of the groundwater recharges follow a more or less smooth curve, with almost no groundwater recharge in the summer season (between 0 and 10 mm) and high groundwater recharge in the winter season (up to 70 mm/month) for both stations. The average groundwater recharge in Amsterdam is slightly higher than in Leeuwarden. The X90 of the monthly groundwater recharge is about 0 most of the year (from February until October). At the station in Leeuwarden, the groundwater recharge from January until April 2006 is normal, with a rather low groundwater recharge in January. No drought occurred during these months. The groundwater recharge for 2006 is low from May until October. The low recharge is normal for May until August. The groundwater recharge in September and October, however, is much lower for 2006 than normal. In fact, the groundwater recharge in these 2 months equals the threshold level. Therefore a drought occurs in the groundwater recharge. The groundwater recharge in November is still below normal, but not as dry as X90, and finally the groundwater recharge in December is even higher than average. At the station in Amsterdam, the groundwater recharge from January until April 2006 is normal. The drought that stroke the summer of 2006 is not enhanced by a low winter groundwater recharge during the previous winter. The period of low groundwater recharge does not take as long as in Leeuwarden. It starts also in May but ends very abruptly in August, with a groundwater recharge that is about 10 times higher as normal for that month. Apparently, low groundwater recharges can quickly change into a high recharge after a large precipitation. September is dry, with a recharge that equals the threshold level (clearly lower than normal). From October onwards, recharge is about normal again. This means that the drought of the summer of 2006 probably has no effect on the summer of 2007 at the station Amsterdam. The absence of a drought in groundwater recharge on clayey soil does not necessarily mean that the 2006 summer drought is not influenced by the previous winter, or that it will have no effect on the summer of 2007. To check this, cumulative groundwater recharges are needed (figures 3.9 and 3.10). It turns out that for both stations, the cumulative monthly winter groundwater recharge before the summer of 2006 is lower than the cumulative average monthly recharge. Therefore, this winter might have influenced the 2006 summer drought. In Amsterdam, the low cumulative groundwater recharge of 2006 is fully compensated by the precipitation in August 2006. The dry summer of 2006 will therefore probably have no effect on the summer of 2007. In Leeuwarden, however, there is still a shortage of 138 mm in groundwater recharge in December 2006. Possibly this shortage will not be recovered before the beginning of the hydrological year of 2007 and will therefore influence the summer of 2007.

26

Figure 3.9. Cumulative groundwater recharge (mm) for clayey soil in Leeuwarden, from January until December, 2006

Figure 3.10. Cumulative groundwater recharge (mm) for clayey soil in Amsterdam, from January until December, 2006

Sandy soil The simulated 30 year averages on monthly groundwater recharges for the investigated sandy soil show the same yearly pattern as the clayey soil, with values that are much alike: between 0 and 10 mm/month in summer and up to 70 mm/month in winter. Average groundwater recharges found at the station in Amsterdam are slightly higher than in Leeuwarden. The X90 of the monthly groundwater recharge is 0 most of the year (from February until October). Also the cumulative groundwater recharges for both stations are similar to the clayey soil (Appendix 7).

Comparison of the clayey an sandy soil It turns out that groundwater recharge (simulated by NUT_DAY) does not seem to depend much on soil type. It depends more on precipitation. A close analysis shows that an almost identical average monthly groundwater recharge was found for both soils at each station. Only during summer time differences of some mm occur, but this is almost negligible and not important for this study, as the main focus is on the winter recharge. The cumulative groundwater recharge also shows this pattern.

27 3.6 Discussion

Generally, it should be noted that in this study, strict thresholds (i.e. X90 and 7 days) are chosen. Choosing less strict thresholds, for instance X75 or 2 days, will lead to identifying more droughts. Especially in situations where the precipitation sums of 2006 are well above the threshold value, but also well below normal.

Precipitation All stations experienced a drought in the precipitation at some moment in the summer of 2006. For most stations, this drought occurred in June and/or July. The ending of the summer drought in the precipitation, is at all stations marked by the high precipitation in August. Precipitation is above X90 for all stations during that month. Even the driest station Leeuwarden receives enough precipitation in August to end the drought. September is dry again at all stations, which means the occurrence of a second, smaller, drought in the precipitation. In October, the stations receive enough precipitation to end that second drought, even in Leeuwarden.

Cumulative potential precipitation deficit The ending of the 2006 summer drought in the cumulative potential precipitation deficit shows a more complex picture than the precipitation. Clearly, a single rainfall event can advance the ending of the summer drought considerably. In 2006, particularly the August rainfall caused the potential precipitation deficit to end at different times at the different stations. The reason for this is the spatial variability of the August precipitation. Choosing a threshold based on the 30 year average cumulative precipitation deficit might be a different approach to determine drought. This might lead to different results for drought in cumulative potential precipitation deficit.

Soil moisture storage in summer The ending of a drought in soil moisture depends obviously on precipitation, but also soil type plays an important role. On a clayey soil, a drought happens sooner than on a sandy soil due to differences in the readily available soil moisture. The selected sandy soil has more readily available soil moisture and is therefore less drought prone. The threshold of readily available soil moisture (critical point, table 2.2), however, could maybe be chosen differently. It is questionable if really a drought occurs once grass cannot evapotranspire in an optimal way. Eventually, once a drought strikes, soil moisture stress can get no lower than a certain level: the soil moisture level at wilting point of that particular soil. This means that if the depth of precipitation equals the soil moisture storage difference between the critical point and wilting point, the drought ends immediately. In this study, the sandy soil will recover sooner from a drought than the clayey soil. Indeed the summer drought in the soil moisture ended the soonest on the sandy soil in Amsterdam, in August. It ended the latest on the clay soil in Leeuwarden, in November. The simulation with NUT_DAY does not consider possible negative side-effects of a long lasting drought on a soil, like hysteresis or water repellency. These features might influence the soil moisture recovering capacity of a soil after a drought (Ritsema and Dekker, 2000). Also, NUT_DAY assumes that all precipitation that falls, infiltrates into the soil immediately. But especially during convective precipitation events, when a lot of precipitation occurs during a short period of time, this might not be the case. Surface runoff might then occur. In August, convective precipitation occurred (Appendix 8). Assuming that

28 possibly not all precipitation in August contributed to soil moisture recovery, the 2006 summer drought in soil moisture might actually have lasted longer for some stations.

Groundwater recharge in winter The monthly averages of groundwater recharge in 2006 almost never dropped below the threshold of X90. Therefore, no groundwater recharge drought occurred during 2006. A lower threshold might lead to different results; more droughts might be identified. However, in most cases, values of the groundwater recharge in 2006 are either normal, or on threshold level. This means that changing the threshold to for instance X75 would not change the groundwater recharge drought a lot. It should be noticed, however, that although no groundwater recharge drought occurred, the winter before the summer of 2006 was slightly dry and therefore might have influenced the 2006 summer drought. Moreover, the dry conditions during the summer of 2006 might in some situations influence the summer of 2007. The groundwater recharge might be more affected if surface runoff in August would have been considered by NUT_DAY. The effect of hysteresis and water repellency on groundwater recharge is also not considered (see ‘soil moisture storage in summer’).

29 30 4. Conclusions and recommendations The study on the 2006 summer drought in the Netherlands has led to the following conclusions:

• Whether a drought ends or not is to some extent dependent on the chosen threshold; • The precipitation a meteorological station receives in the early summer (e.g. April or May) can influence the maximum cumulative potential precipitation deficit per station. Hence it influences the ending of the drought; • During summer time, precipitation can have a convective nature. Therefore the precipitation that a meteorological station receives during summer time can vary significantly from location to location. In 2006 this leads to different moments of ending of the drought in the cumulative potential precipitation deficit; • The occurrence of a drought in soil moisture is strongly dependent on soil type. Which have different available soil moisture storages. A soil with a high readily available soil moisture storage is less sensitive to drought. Different soils will likely have different amounts of soil moisture storage between wilting point and critical point (table 2.2). If during a drought a soil is at wilting point, a soil with a lower amount of soil moisture storage between critical and wilting point will sooner recover once precipitation occurs; • The occurrence of a drought in groundwater recharge does not seem to depend too much on soil type, more on precipitation; • The 2006 summer drought might have been influenced by the low groundwater recharge in the winter before. The dry conditions of the 2006 summer might for some locations influence the summer of 2007; • The selected meteorological stations are biased to the North and West of the Netherlands and are therefore not representative for the Netherlands in general; • NUT_DAY does not take soil altering effects of drought into consideration. This might influence the results of this study. The ending of droughts in soil moisture and groundwater recharge might be different because of these effects; • NUT_DAY does not take the convective nature that summer precipitation can have into consideration. NUT_DAY assumes that surface runoff never takes place. This might influence the results of this study. Droughts according to soil moisture and groundwater recharge might take longer to recover in reality because of this assumption.

The study on the 2006 summer drought in the Netherlands has led to the following recommendations:

• More meteorological stations could have been investigated, to get a more detailed and less biased spatial overview. This could lead to an explanation of the deviating cumulative potential evapotranspiration for Leeuwarden. Moreover, the results could support the cumulative precipitation deficit maps from KNMI more; • Other thresholds on precipitation, cumulative potential precipitation deficit, soil moisture and groundwater recharge could be studied, to investigate their importance in identifying droughts and drought endings; • The Slob algorithm and global radiation data by KNMI could better be used to calculate sunshine duration. Calculating global radiation from sunshine duration and

31 optimizing the found values, like was done in this study, probably leads to larger errors; • The behaviour of convective precipitation could be investigated, to obtain more knowledge on how much precipitation actually discharges as surface runoff during such a precipitation event. NUT_DAY could then be adapted; • Grass is the only crop that is investigated in this study. Possibly, investigating other crops might lead to different outcomes on drought occurrence and ending, at least for drought in soil moisture and groundwater recharge.

32 References Allen, R.G., Pereira, L.S., Raes, D., Smith, M. (1998) Crop evapotranspiration. Guidelines for computing crop water requirements. Irrigation and Drainage Paper 56, FAO, Rome, Italy.

Dam, J.C. van, Feddes, R.A., Witte, J.P.M. (2005) Soil Physics and Agrohydrology. Wageningen University, Wageningen.

Feddes, R.A. (1987) Crop factors in relation to Makkink reference crop evapotranspiration . In: J.C. Hooghart (red) Verslagen en Mededelingen nr 39, Cornmissie voor Hydrologisch Onderzoek TNO, Den Haag, pg. 3345.

Hisdal, H., Tallaksen, L.M., Clausen, B., Peters, E., Gustard, A. (2004) Hydrological Drought Characteristics. In: Hydrological drought: processes and estimation methods for streamflow and groundwater. Development in Waterscience, 48, Elsevier pg. 139-198.

Huisman, P., Cramer, W., van Ee, G., Hooghout, J.C., Salz, H., Zuidema, F.C. (1998) Water in the Netherlands. Netherlands Hydrological Society (NHV), Delft.

KNMI (2006) Monthly report of evapotranspiration and precipitation 2006 (in Dutch).

Lanen, H.A.J. van, Weerts, A.H., Kroon, T., Dijksma, R. (1996) Estimation of groundwater recharge in areas with deep groundwater tables using transient groundwater flow modelling. Proc. Int. Conf. on ‘Calibration and Reliability of Groundwater Modelling’, September 2006, Golden, USA, pg 307 316.

Lanen, H.A.J. van. (2006) 2006 Drought in the Netherlands at the end? Web report, European Drought Centre.

Locher, W.P. and de Bakker, H. (1990) Soils of the Netherlands; part 1. General Soil Science (in Dutch). Malmberg, den Bosch.

Makkink, G.F. (1957) Testing the Penman formula by means of lysimeters. J. Int. Water Eng., 11:277-288.

Peters E., van Lanen, H.A.J., Alvarez J. and Bradford, R.B.B. (2001) Groundwater droughts: evaluation of temporal variability of recharge in three European groundwater catchments. ARIDE Technical report no.11, Wageningen.

Ritsema, C.J., and Dekker, L.W. (2000) Water repellency in soils. Special issue Journal of Hydrology, pg 231- 232.

RIZA, HKV, Arcadis, KIWA, Korbee and Hovelynck, Klopstra, D., Versteeg, R., Kroon, T. (2005) Drought study the Netherlands; nature, severity and magnitude of water shortage in the Netherlands (in Dutch). RIZA rapport 2005.016. Ministerie van Verkeer en Waterstaat, Directoraat Generaal Water.

Slob, W. H., and Monna, W. A. A. (1991). Determination of direct and diffuse radiation and of sunshine duration from 10-minute values of global radiation (in Dutch). Technische rapporten, TR-136. De Bilt, Ministerie van Verkeer en Waterstaat, Koninklijk Nederlands Meteorologisch Instituut.

Tallaksen, L.M. and van Lanen, H.A.J. (2004) Hydrological drought: processes and estimation methods for streamflow and groundwater. Development in Waterscience, 48, Elsevier, pg 3-17

Tannehill, I.R. (1947) Drought and Its Causes and Effects. Princeton University Press

33

34 Appendix

35 Appendix 1

until July 23 2006 Figure A1(d). Cumulative potential 1 April from deficit; precipitation potential Cumulative A1(h). Figure 1 until April from deficit; precipitation 2006 27, July

Figure A1(c). Cumulative potential potential Cumulative A1(c). Figure 1 April from deficit; precipitation 2006 20, July until Figure A1(g). Cumulativepotential 1 April from deficit; precipitation 2006 26, July until

potential A1(b). Cumulative Figure 1 April from deficit; precipitation 19, 2006 July until potential A1(f). Cumulative Figure 1 April from deficit; precipitation 25, 2006 July until

Figure A1(a). Cumulative potential potential Cumulative A1(a). Figure 1 April from deficit; precipitation 2006 17, July until potential Cumulative A1(e). Figure 1 April from deficit; precipitation 2006 24, July until

36

potential A1(l). Cumulative Figure 1 April from deficit; precipitation 2006 31, July until Figure A1(p). Cumulative potential potential Cumulative A1(p). Figure 1 April from deficit; precipitation 8, 2006 August until

potential Cumulative A1(k). Figure 1 April from deficit; precipitation 2006 30, July until potential Cumulative A1(o). Figure 1 April from deficit; precipitation 2006 7, August until

Figure A1(j). Cumulativepotential from 1 April deficit; precipitation 2006 29, July until potential Cumulative A1(n). Figure from 1 April deficit; precipitation until August 5, 2006

Figure A1(i). Cumulative potential potential A1(i). Cumulative Figure 1 April from deficit; precipitation 2006 28, July until potential A1(m). Cumulative Figure 1 April from deficit; precipitation 4, 2006 August until

37

potential A1(t). Cumulative Figure 1 April from deficit; precipitation 12, 2006 August until potential A1(x). Figure Cumulative 1 April from deficit; precipitation 17, 2006 August until

Figure A1(w). Cumulative potential potential Cumulative A1(w). Figure 1 April from deficit; precipitation 2006 16, August until Figure A1(s). Cumulative potential 1 April from deficit; precipitation 2006 11, August until

Figure A1(v). Cumulative potential potential Cumulative A1(v). Figure 1 April from deficit; precipitation 2006 15, August until Figure A1(r). Cumulativepotential 1 April from deficit; precipitation 2006 10, August until

Figure A1(u). Cumulative potential potential Cumulative A1(u). Figure 1 April from deficit; precipitation 2006 14, August until potential Cumulative A1(q). Figure 1 April from deficit; precipitation 9, 2006 August until

38

potential A1(B). Figure Cumulative 1 April from deficit; precipitation 26, 2006 August until potential A1(F). Figure Cumulative 1 until April from deficit; precipitation 2006 20, September

Figure A1(E). Cumulative potential potential A1(E). Cumulative Figure 1 April from deficit; precipitation 2006 14, September until potential A1(A). Cumulative Figure 1 April from deficit; precipitation 2006 21, August until

Figure A1(D). Cumulative potential potential Cumulative A1(D). Figure 1 April from deficit; precipitation 8, 2006 September until Figure A1(z). Cumulativepotential 1 April until from deficit; precipitation 2006 19, August

Figure A1(C). Cumulative potential potential A1(C). Cumulative Figure 1 April from deficit; precipitation 2006 4, September until potential Cumulative A1(y). Figure 1 April from deficit; precipitation 2006 18, August until

39

potential A1(G). Cumulative Figure 1 April from deficit; precipitation 28, 2006 September until

40 Appendix 2

Figure A2.1 Precipitation in Groningen in Groningen Precipitation Figure A2.1 Helder in den Precipitation A2.2 Figure in Twenthe Precipitation Figure A2.3 in Leeuwarden Precipitation A2.4 Figure

41

Figure A2.8. Precipitation Maastricht Maastricht Precipitation A2.8. Figure

Amsterdam in Precipitation A2.6. Figure

Figure A2.7. Precipitation Vlissingen Vlissingen Precipitation A2.7. Figure Figure A2.5. Precipitation Rotterdam Rotterdam Precipitation A2.5. Figure

42

Appendix 3

Figure A3.1. Cumulative evaporation from May 1 until July 31, de Bilt

Figure A3.2. Cumulative evaporation from May 1 until July 31, den Helder

43 Appendix 4

April 1 from Twenthe, for deficit precipitation potential Cumulative A4.4. Figure 2006 10, December until Figure A4.2. Cumulative potential precipitation deficit for Leeuwarden, from from Leeuwarden, for deficit precipitation potential Cumulative A4.2. Figure 2007 16, January 1 until April

from Helder, den for deficit precipitation potential Cumulative A4.1. Figure 2006 17, November until April 1 Figure A4.3. Cumulative potential precipitation deficit for Groningen, from from Groningen, for deficit precipitation potential Cumulative A4.3. Figure 2006 29, December until 1 April

44

Figure A4.8. Cumulative potential precipitation deficit for Maastricht, from from Maastricht, for deficit precipitation potential Cumulative A4.8. Figure 2006 6, December until 1 April from Amsterdam, for deficit precipitation potential Cumulative A4.6. Figure 3) October at ends deficit minor (second, 2006 28, August until 1 April

Figure A4.7. Cumulative potential precipitation deficit for Vlissingen, from from Vlissingen, for deficit precipitation potential Cumulative A4.7. Figure 2006 11, December until 1 April from Rotterdam, for deficit precipitation potential Cumulative A4.5. Figure 2006 21, October 1 until April

45 Appendix 5

Figure A5.4. Soil moisture storage (mm) sandy soil Amsterdam, from Amsterdam, soil sandy (mm) storage Soil moisture A5.4. Figure 2006 December, until January from soil clayey Amsterdam, (mm) storage Soil moisture A5.2. Figure January untilDecember, 2006

006 2 ,

r

ecembe

D

until

y

r

Janua Figure A5.3. Soil moisture storage (mm) sandy soil Leeuwarden, from from Leeuwarden, soil sandy (mm) storage moisture Soil A5.3. Figure from soil Leeuwarden, clayey (mm) storage Soil moisture A5.1. Figure 2006 December, until January

46 Appendix 6

Figure A6.4. Soil moisture storage (mm) for sandy soil Amsterdam from April 1, April from Amsterdam soil sandy for (mm) storage moisture Soil A6.4. Figure the at storage moisture soil the represents (y=0 31, 2006. December until 2006 point) critical April from soil Amsterdam for clayey (mm) storage Soil moisture A6.2. Figure at storage moisture soil the represents (y=0 2006. 31, December until 2006 1, point) the critical

Figure A6.1. Soil moisture storage (mm) for clayey soil Leeuwarden from from soil Leeuwarden clayey for (mm) storage Soil moisture A6.1. Figure moisture soil the represents (y=0 2006. 31, December until 2006 1, April point) at the critical storage April from soil Leeuwarden sandy for (mm) storage Soil moisture A6.3. Figure at storage moisture the soil represents (y=0 2006. 31, December until 2006 1, point) critical the

47 Appendix 7

Figure A7.2. Groundwater recharge (mm) for sandy soil in Amsterdam, 2006 December, until January from

in soil for sandy (mm) recharge groundwater Cumulative FigureA7.4. 2006 December, until January from Amsterdam,

Figure A7.1. Groundwater recharge (mm) for sandysoilin Leeuwarden, 2006 until December, January from in soil for sandy (mm) recharge groundwater Cumulative A7.3. Figure 2006 December, until January from Leeuwarden,

48 Appendix 8

Figure A8.1. Daily precipitation sums de Bilt, August Figure A8.2. Daily precipitation sums in 2006. the Netherlands, August 1, 2006. Blue is amount of precipitation (mm), red is precipitation duration (h).

Figure A8.2. Daily precipitation sums in Figure A8.2. Daily precipitation sums in the Netherlands, August 11, 2006. the Netherlands, August 21, 2006.

49