estimating the effects of climate change on groundwater ... the effects of climate... · abstract...

Hydrological Sciences–Journal–des Sciences Hydrologiques, 54(4) August 2009 Special issue: Groundwater and Climate in Africa

* Now at: UNESCO-IHE, Institute of Water Education, Delft, The Netherlands.

713

Estimating the effects of climate change on groundwater recharge and baseflow in the upper Ssezibwa catchment, Uganda PHILIP M. NYENJE1* & OKKE BATELAAN2,3

1 Interuniversity Programme in Water Resources Engineering, Katholieke Universiteit Leuven & Vrije Universiteit Brussel, Belgium [email protected]

2 Department of Hydrology and Hydraulic Engineering, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels, Belgium

3 Department of Earth and Environmental Sciences, Katholieke Universiteit Leuven, Celestijnenlaan 200E, B-3001 Heverlee, Belgium Abstract The effects of climate change on groundwater recharge and baseflow in the upper Ssezibwa catch-ment, Uganda, are investigated. The study first examines historical data, which indeed reveal evidence of climate change based on trends observed in temperature and discharge. For the climate change study, the statistical downscaling model (SDSM) is used to downscale future climate change scenarios, which were obtained from the UK HadCM3 climate model. The downscaled climate is used as input to the WetSpa hydrological model, a physically-distributed rainfall–runoff model, which was used to simulate the resulting hydrological changes. Downscaled climate shows an increase in precipitation in the wet seasons (March–May; October–December) ranging from 30% in the 2020s to over 100% in the 2080s. The corresponding rise in temperature ranges between 1 and 4°C. These changes are shown to give rise to intensification of the hydrological cycle. The mean annual daily baseflow for the current period of 157 mm/year (69% of dis-charge), is expected to increase by 20–80% between the 2020s and 2080s. The corresponding increase in recharge ranges from 20 to 100% from the current 245mm/year. The findings provide a basis for further research in the downscaling of climate data and sensitivity analysis for simulated hydrological changes in the catchment. Key words climate change; groundwater recharge; baseflow; downscaling; GCM; WetSpa

Estimation des effets du changement climatique sur la recharge en eaux souterraines et l’écoulement de base dans le bassin versant de Ssezibwa supérieur, Ouganda Résumé Les effets du changement climatique sur la recharge en eaux souterraines et sur l’écoulement de base sont étudiés dans le bassin versant de Ssezibwa supérieur, en Ouganda. L’étude examine tout d’abord les données historiques, qui mettent effectivement en évidence un changement climatique sur la base des tendances observées de température et de débit. Pour l’étude du changement climatique, le modèle statistique de descente d’échelle (SDSM) est utilisé pour régionaliser des scénarios futurs de changement climatique, obtenus avec le modèle climatique Britannique HadCM3. Le climat régionalisé est utilisé en entrée du modèle hydrologique WetSpa, un modèle pluie–débit distribué physiquement, pour simuler les changements hydrologiques induits. Le climat régionalisé montre une augmentation des précipitations pendant les saisons humides (Mars–Mai; Octobre–Décembre) allant de 30% dans les années 2020 à plus de 100% dans les années 2080. L’augmentation de température correspondante va de 1 à 4°C. Ces changements se révèlent conduire à une intensification du cycle hydrologique. L’écoulement de base journalier moyen annuel de la période actuelle de 157 mm/an (69% de l’écoulement) devrait augmenter de 20–80% entre les décennies 2020 et 2080. L’augmentation correspondante de la recharge va de 20 à 100% à partir des 245 mm/an actuels. Les résultats fournissent une base à de futures recherches en termes de régionalisation par descente d’échelle de données climatiques et d’analyse de sensibilité de la modélisation des changements hydrologiques dans le bassin versant. Mots clefs changement climatique; recharge en eaux souterraines; écoulement de base; descente d’échelle; MCG; WetSpa INTRODUCTION One of the most severe consequences of climate change will be the alteration of the hydrological cycle, which will impact on the quantity and quality of regional water resources (Gleick & Adams, 2000; Xu, 2000). According to the Intergovernmental Panel on Climate Change (IPCC, 2001), African countries are more vulnerable to these changes due to lack of institutional capacity and

Open for discussion until 1 February 2010 Copyright © 2009 IAHS Press

Philip. M. Nyenje & Okke Batelaan

Copyright © 2009 IAHS Press

714

economic development. The IPCC (2001, 2007) reports that overall, Africa has warmed by 0.7°C over the 20th century while rainfalls have increased by 5–10%. However, future projections indicate that temperatures will be 1.5°C warmer by 2050, while precipitation will increase in equatorial Africa and decrease by about 10% in northern Africa (IPCC, 2007). Joubert & Hewitson (1997) report similar predictions where the mean summer rainfall in Africa is expected to generally decrease between 10 and 20% by the end of the 21st century. These changes will increase evaporation and cause more intense rain storms, severely affecting the availability, quantity and quality of water (IPCC, 2001; Phoon et al., 2004). Heavy rainfalls in the wet seasons are expected to damage agriculture and engineering structures as a result of increased runoff (Arnell, 1999). In the East African region, the increase in runoff is generally expected to reach 10–20% by the 2050s (Milly, 2005). However, significant reductions in runoff are expected in summer for most of Africa (Arnell, 1999). Current research has mainly focused on surface water and very little is known about the potential impacts on groundwater. Yet, the impacts on groundwater are far reaching and need to be investigated, especially in Africa, where most people rely on groundwater as a source of potable water and for other domestic uses (Taylor & Howard, 1996; Kulabako et al., 2007). In Uganda, for example, 61% of the country’s safe water is supplied by groundwater, mainly through use of springs around Lake Victoria and southern parts Uganda, and boreholes in northern Uganda (Tindimugaya, 2006). Groundwater is a vital water resource and awareness needs to be raised on its vulnerability to overexploitation, pollution and, most importantly, climate change. However, there are still few research studies that have explored the effects of climate change on groundwater. Therefore, we attempt in this study to investigate the effects of climate change on groundwater in a catchment in Uganda. The study further tests the downscaling of climate change projections from global climate models (GCMs), which has received relatively little attention in Africa (Joubert & Hewitson, 1997). STUDY AREA The upper Ssezibwa catchment (Fig. 1) is located in central Uganda and covers an area of about 175 km2. It drains into Lake Kyoga and is bordered by the Lake Victoria basin to the south. The catchment is gauged at Ssezibwa falls (0°21′N; 32°52′E), which is just upstream of Ssezibwa River, because the lower part is too swampy to establish a meaningful station. The mean daily flow is reported to range between 1 to 3 m3/s with a maximum flow of about 10 m3/s (Otim, 2005). The catchment elevation varies from 1122 m in the north to 1353 m in the south, with a mean slope of 6.9%. Agriculture is the predominant land use (about 63%), while the remaining catch-ment is covered by evergreen broad-leaf forest. The soil texture is mainly sandy clay-loam. The region is characterized by a tropical humid climate influenced by Lake Kyoga and Lake Victoria (NEMA, 1997). The area experiences two rain seasons (March–May and September–December) with a mean annual rainfall of 1600–2000 mm. The annual evaporation is about 1472 mm, which renders the area a rainfall surplus zone. The mean annual maximum temperature is 26°C, while the minimum is 16°C. The catchment is generally underlain by undifferentiated gneisses associated with rocks of the Precambrian era. Younger rock formations of mainly sedimentary origin, typically with a thickness of 30 m or less, generally overlay the old gneisses (Taylor & Howard, 1996; Morris et al., 2003). According to Taylor & Howard (1996, 2000), these materials are the origin of the valuable, often overlooked, aquifer in the region. DATA AVAILABILITY Data used in this research included 90-m-resolution grid maps of elevation, land use and soil, as well as daily precipitation and potential evapotranspiration data (PET), daily discharge and daily large-scale GCM predictors, which are used to predict future climate changes. The elevation grid map at 90-m resolution was obtained from the elevation database of the Shuttle Radar

Estimating the effects of climate change on groundwater recharge and baseflow


715

Fig. 1 Map showing the location of the study area within Uganda and the meteorological and gauge stations.

Topographic Mission (US Geological Survey: http://srtm.usgs.gov). The land-use and drainage maps were obtained from the FAO Africover database (http://www.africover.org/index.htm). This mapping is largely based on satellite imagery data with a scale of 1:200 000. Long series of daily meteorological data were available for six weather stations around the catchment and were obtained from the Uganda Meteorological Department. Table 1 lists the available stations. The weather variables include: precipitation, minimum and maximum tempera-ture, wind speed and relative humidity. Three stations (Namulonge, Jinja and Kampala) were used for PET, while the other three (Lugazi, Kivuvu and Mitono) were used for precipitation. The latter are located close to the catchment making them valuable for the precipitation interpolation, since rainfall is highly variable in the catchment. The PET series were calculated from the available meteorological data using the Penman-Monteith formula. Discharge records spanning from 1 January 1960 to 1 August 2006 were obtained from the Directorate of Water Development (DWD), Uganda. As shown in Table 1, these data contained several gaps and therefore the study period was limited from 1997 to 2004. Gaps in data were filled by estimation from the closest stations having available data for that period. The precipitation stations considered were first checked for consistency using the double mass curve method. Table 1 Summary of meteorological and hydrological stations for the catchment. ID Name of station Location From To % Missing (years) Meteorological stations: 1 Namulonge 0.53°N–32.61°N 1968 2005 10 2 Jinja 0.45°N–33.18°N 1970 2004 7 3 Kampala 0.31°N–32.61°N 1993 2005 0 4 Lugazi 0.42°N–32.93°E 1979 2004 20 5 Kivuvu 0.40°N–32.93°E 1979 2004 30 6 Moniko 0.42°N–32.90°E 1980 2004 40 Stream gauging station: - Ssezibwa Falls 0.38°N–32.87°E 1960 2006 50



716

METHODOLOGY Groundwater resources can be related to climate change in terms of the nature of recharge and the interaction between groundwater and surface water (Gleick & Adams, 2000; Loaiciga, 2003). Hydrological models are therefore usually used to quantify recharge, which is then used as input to groundwater models to simulate changes in groundwater levels (Loaiciga, 2003; Scibek & Allen, 2006; Jyrkama et al., 2007; Woldeamlak et al., 2007). The assessment involves comparing the model outputs for specified climate change scenarios to the current climate or baseline climate, which is taken as a 30-year period from 1961 to 1990 (Phoon et al., 2004; Mitchell, 2005). In this study, the effects of climate change on the groundwater system of the catchment are assessed based on changes in the recharge and baseflow. Both are key components for assessing ground-water resources (Lee et al., 2002). The recharge and baseflow are estimated using the WetSpa (Water and Energy Transfer between Soil, Plants and Atmosphere) hydrological model (Liu et al., 2004). Future climate change scenarios are based on global circulation model (GCM) outputs obtained from the UK Hadley Centre climate model, HadCM3. However, the typical grid size (2.5° by 3.75°) is too coarse for a hydrological impact analysis at a watershed scale (Wigley et al., 1990; O’Hare et al., 2005). This study therefore employs the Statistical Downscaling Model (SDSM) to convert the grid GCM outputs to finer resolutions, which are suitable for hydrological analysis at the catchment level. Thus, the methodology was split into four steps: (a) trend analysis; (b) distributed hydrological modelling; (c) statistical downscaling of climate change scenarios; and (d) hydrological response analysis. Trend analysis A trend analysis was performed to investigate whether climate has changed in the catchment in the past years. The analysis was performed on the discharge, precipitation and mean temperature using the Trend Testing Tool (Willems, 2006). Jinja station was used for the analysis of precipitation and temperature because it was the nearest station and had the most continuous and longest time series. Trend analyses require long data series, so the period 1960–2006 was used. To minimise gaps and ensure an independent input series, the analysis for precipitation and temperature was performed on monthly series derived from the daily values, while the analysis for discharge was based on peak over threshold (POT) or partial duration series, which were extracted from the daily series using the WETSPRO program (Willems, 2009). The POT series contains an array of all inde-pendent peak flood values exceeding a selected threshold. The threshold used was based on the method in which two successive peaks are considered independent when the time, p, between them is longer than the recession constant (10 days as determined by WETSPRO), and when the mini-mum discharge between the two peaks is smaller than a fraction, f, of the peak discharge. For this study, the following parameters were found appropriate: p = 10 d, f = 0.5. The results of the ana-lysis are summarized in Fig. 2. Two periods are used for the analysis; 1960–1990 and 1990–2006, because trends are dependent on the period analysed (Jónsdóttir et al., 2005). As shown in Fig. 2, trends, especially in discharge, generally become apparent in the second period (1990–2006). However, trends are not significant for precipitation. Discharge series show decreasing trends from 1990, estimated at 50% per decade. In contrast, temperatures are shown to have been gradually increasing since 1960 and the trend is approximated at 2% increase per decade (Fig. 2(c)). Distributed hydrological modelling WetSpa is a physically-based distributed rainfall–runoff model (Liu et al., 2004, 2005). Figure 3 shows schematically the model structure at the grid cell level. The water balance of the root zone is considered a key factor controlling runoff, interflow and groundwater recharge and is calculated for each grid cell as:

D P I S E F RtθΔ = − − − − −

Δ (1)

where D is the root zone depth; Δθ is the change in soil moisture content; Δt is the time interval;



717

0

5

10

15

20

25

0 50 100 150 200POT Counts since 1960

Q (m

3 /s)

observationsmeanRegression

0

5

10

15

20

25

0 20 40 60 8

POT counts since 1994

Q (m

3 /s)

.

0

observationsmeanRegression

(a) Discharge (1960–2006) (b) Discharge (1994–2006)

18

20

22

24

26

0 100 200 300 months since 1960

Mon

tly te

mp.

(°C

)

observations mean Regression

0

100

200

300

400

0 50 100 150 200months since 1990

Mon

thly

pre

cip.

(mm

) . . observations

meanRegression

(c) Temperature (1960–2004) (d) Precipitation (1990–2004)

Fig. 2 Trend analysis plots: (a), (b) discharge, (c) temperature and (d) precipitation.

Fig. 3 Structure of the WetSpa hydrological model.

P is the precipitation; I is the initial abstraction (interception and depression losses); S is the surface runoff; E is the actual evapotranspiration; F is the interflow; and R is the percolation out of the root zone. The percolation out of the root zone recharges the groundwater storage (SG), which then contributes to groundwater discharge forming the baseflow of a stream hydrograph (Liu et al., 2004). Recharge is estimated using equation (2) based on the Brooks & Corey relationship between hydraulic conductivity and effective saturation (Brooks & Corey, 1966):

(2 3 ) /B B

rs

s r

R K θ θθ θ

+⎛ ⎞−= ⎜ ⎟−⎝ ⎠

(2)



718

and baseflow simulation is performed at the sub-catchment scale using a lumped linear reservoir approach:

Qg = Cg (SG/1000) (3) where R is the recharge or percolation; Ks is the saturated soil hydraulic conductivity; θr is the residual moisture content; B is the pore size distribution index; Qg(t) is the baseflow; SG(t) is the groundwater storage; and Cg is the groundwater flow recession constant. The main inputs to the model are: precipitation, PET, discharge series and 90-m GIS grid maps of elevation, land use and soil. The main outputs are: river flow hydrographs and spatially-distributed hydrological characteristics, such as soil moisture, infiltration rates and groundwater recharge. The model is coupled with ArcView to enable pre-processing of GIS data, such as delineation of the spatial inputs of land use, soil and elevation, and derivation of the model’s spatial parameters, such as Manning’s and runoff coefficients, time of travel, flow velocity and surface routing parameters. In setting up the model, the catchment was divided into 110 sub-catchments, the maximum needed to define a stream in the catchment. The average runoff coefficient was obtained as 0.46 and the average flow time as 12.8 h with a maximum of 41 h. To run the model, the three stations Lugazi, Kivuvu and Moniko were used for precipitation input, while Namulonge, Jinja and Kampala were used for PET input (cf. Table 1). The distribution of precipitation and PET in the catchment was determined using Thiessen polygons created for all stations with ArcView. Each grid is then identified with the station number and value of the corresponding polygon. The model was calibrated by manually adjusting the model’s global parameters using daily flows from 1 January 1997 to 31 December 1999. The automatic parameter estimator (PEST; Doherty, 2002) was also used to fine-tune the parameters. The results of the calibration were validated using an independent data set from 1 January 2002 to 31 December 2004. For the semi-distributed model, the calibration resulted in a Nash-Sutcliffe efficiency of 0.629, while the efficiency for the validation period was 0.798. Hence the calibration results are acceptable. A graphical comparison of the simulated and observed flows for the validation period is shown in Fig. 4.

0

4

8

12

16

20

1/1/

02

3/1/

02

5/1/

02

7/1/

02

9/1/

02

11/1

/02

1/1/

03

3/1/

03

5/1/

03

7/1/

03

9/1/

03

11/1

/03

1/1/

04

3/1/

04

5/1/

04

7/1/

04

9/1/

04

11/1

/04

Dis

char

ge (m

3 /s)

0

40

80

120

160

200

Pre

cip.

(mm

)

Rainfall (mm)QcalculatedQmeasured

Fig. 4 Validation results for the WetSpa model for the period 2002–2004.

Statistical downscaling of climate change scenarios A statistical downscaling model (SDSM; Wilby et al., 2002) was used to downscale daily GCM climate projections of the catchment for use in the WetSpa model. Downscaled data are more reliable for hydrological analysis at the catchment level as compared to GCM data, which often distort patterns of rainfall variability (Karl et al., 1990). The SDSM model calculates relationships, which relate large-scale GCM predictor variables, such as geopotential heights and humidity, to



719

local-scale meteorological series (predictants). The relationships are then used to downscale future climate change scenarios using GCM predictor variables. The predictants considered were the locally observed daily temperature and precipitation series, averaged for all the stations in the catchment, from the period 1961–2001. The GCM predictors for the current period (1961–2001) and for the future periods of the 2020s (2010–2039), 2050s (2040–2069) and 2080s (2070–2099) were obtained from the UK Hadley Centre’s global circulation model, HadCM3, using the location 0°N, 33.75°E. Two emission scenarios were considered: IPCC SRES scenario A2 (high) and scenario B2 (low). Emission scenarios predict the emission of greenhouse gases, which are the main driving factors of the GCM predictions (Dibike & Coulibaly, 2004). Scenario A2 describes a heterogeneous world with a continuously increasing population and economic growth, and scenario B2 describes a lower population and intermediate levels of economic development. Observed large-scale predictors for the historical period are derived from the NCEP (National Centre for Environmental Protection) data set, and are used along with the locally observed meteorological data to calibrate and validate the SDSM model. The NCEP data set is a re-analysis of the current state of the global atmosphere derived using historical data from 1948 onwards and is essentially free of inhomogeneities (Kistler et al., 2001). The most appropriate predictors for use in the downscaling process are identified during the screening of predictor variables. This is achieved using linear correlation analysis and scatter plots between the NCEP predictors and predictant variables for the current period 1961–2001. Potential predictors are those that give the highest correlation with the predictants. Table 2 summarizes the available potential predictors identified for the catchment. The correlations were generally poor, especially for precipitation. This was attributed to gaps in the predictants (22% for precipitation and 30% for temperature), the variability of rainfall and the limited predictors for the catchment, as the HadCM3 model does not provide surface flow predictors for areas along the Equator, where the catchment is situated. Table 2 Available potential NCEP predictors for the catchment. No. Precipitation predictors Temperature predictors 1 p500 p500 2 r850, r500 r850 3 rhum rhum, shum 4 shum Temp p500: 500 hPa geopotential height; r500: relative humidity at 500 hPa height; r850: relative humidity at 850 hPa height; shum: near surface specific humidity; rhum: near surface relative humidity; temp: mean temperature. The selected predictors were used to calibrate the temperature and precipitation SDSM models for the period 1961–1990. For the calibration of precipitation, an event threshold of 0.03 mm, a variance inflation of 13 and a bias of 0.8 were found appropriate. The simulations generated by the SDSM consisted of 20 ensembles. The ensemble mean was used for the analysis. The downscaled temperature was analysed for the mean and variance, while precipitation was analysed for the mean, variance, percentage wet days and dry series length. The period 1991–2001 was used to validate the downscaled data. In Fig. 5, the downscaled NCEP data are verified with the locally observed data (predictants) to identify how well the downscaled climate represents that of the catchment. As shown in Fig. 5, a good match was obtained. Results of downscaled climate The calibrated temperature and precipitation SDSM models were used to downscale the HadCM3 GCM data for the baseline/current period (1961–1990) and for the three future periods; the 2020s (2010–2039), the 2050s (2040–2069), and the 2080s (2070–2099). Figure 6(a) and (b) shows the comparison of mean daily values of the downscaled current and future precipitation for the two emission scenarios. These results indicate that precipitation will generally increase, with the greatest increase expected in the wet seasons (March–May;



720

(b)

15

17

19

21

23

25

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Mea

n da

ily te

mp

(°C

) .

Observed (NCEP)

Observed (LOCALLY)

(a)

0

2

4

6

8

10

12

14

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Mea

n da

ily p

reci

p (m

m)

. Observed (NCEP)

Observed (LOCALLY)

Fig. 5 Verification of downscaled NCEP climate for the present period, 1961–1990 for: (a) monthly mean daily precipitation, and (b) monthly mean daily temperature.

High Emission (A2)

0

5

10

15

20

25

30

35

Jan

Feb

Mar Ap

r

May Jun

Jul

Aug

Sep

Oct

Nov

Dec

Mea

n da

ily p

reci

p. (m

m)

.

Current 2020's

2050's 2080's

High Emission (A2)

20

22

24

26

28

Jan

Feb

Mar

Apr

May Jun

Jul

Aug

Sep Oct

Nov

Dec

Mea

n da

ily te

mp.

(°C

) . Current 2020's

2050's 2080's

Low Emission (B2)

0

5

10

15

20

25

30

35

Jan

Feb

Mar

Apr

May Jun

Jul

Aug

Sep

Oct

Nov

Dec

Mea

n da

ily p

reci

p. (m

m)

.Current 2020's2050's 2080's

Low Emission (B2)

20

22

24

26

28

Jan

Feb

Mar Ap

r

May Jun

Jul

Aug

Sep

Oct

Nov

Dec

Mea

n da

ily te

mp.

(°C

) . Current 2020's

2050's 2080's

(a)(a) (b)(b)

(c)(c) (d)(d) Fig. 6 Comparison of downscaled climate data for the present and future periods: (a) and (b) down-scaled precipitation for high- and low- emission scenarios, respectively; (c) and (d) downscaled temperature for high- and low- emission scenarios, respectively.

October–December) by over 30% in the 2020s, over 80% in the 2050s and over 100% in the 2080s. Figure 6(c) and (d) compares the results of the downscaled temperature. These results suggest that the months of May–September (the dry season) are expected to experience the greatest increases in temperature, in the range of 1–3ºC, between the 2020s and the 2080s for intense greenhouse scenario A2, and 1–2°C for the low-emission B2 scenario. Mean daily temperatures are expected to reach values ranging between 25 and 26ºC in the 2080s. Generally, it can be seen that the low-emission simulations (B2) are of lower magnitude than the high-emission simulations (A2).



721

Hydrological response simulations

The downscaled time series of daily precipitation and PET corresponding to current and future climate change scenarios from 1961 to 2099 were input into the calibrated semi-distributed WetSpa model for simulation of the discharge and baseflow responses. As it is not possible to use the multi-variable Penman-Monteith method to calculate future evaporation rates, due to data limitations, the PET series was calculated from the downscaled temperature (Tm) using the temperature-based Blaney-Criddle equation (Blaney & Criddle, 1950):

)13.846.0(PET += mTKp (4)

where p is the monthly or daily percentage of daytime hours; and K is the monthly consumptive use coefficient, depending on vegetation type, location and season. At the equator, p = 0.27. The value of K was calculated by comparing the calculated PET with the observed PET for the current period 1961–1990. A value of K = 0.75 was found appropriate. The distributed recharge is calculated by running the fully-distributed WetSpa model for the current period and the future periods. SIMULATION RESULTS AND DISCUSSION

The simulated discharge and baseflows were analysed on a mean monthly basis. Figure 7 shows the simulated flows for the current (1961–1990) and the future periods (2020s, 2050s and 2080s) for both the emission scenarios A2 (high) and B2 (low). These results show that there is generally an increase in both the discharge and baseflow between November and June, while between July and October, little or no change is expected (Fig. 7). However, greater increases are expected in runoff than in baseflow. These changes become more pronounced after the 2050s. The highest flows (ranging between 200 and 350 m3/s) were simulated to occur in the 2080s between the months of November and December. Examining Fig. 6 and Fig. 7 suggests that the high flows are more correlated to changes in precipitation than temperature. Figure 7(e) and (f) shows the varia-tion in baseflow compared to the total river flow. The results suggest that almost 100% of the river flow in the dry season (July–September) is contributed by baseflow for all scenarios. In the wet season, the amount ranges from 20 to 80% between the 2020s and 2080s. On average, the mean annual daily discharge for the present period was estimated by the model at 1.47 m3/s. This value is close to the mean discharge reported for the Ssezibwa River by Otim (2005). For the high-emission scenario, this value was simulated to increase by 38% in the 2020s, 60% in the 2050s and over 100% in the 2080s. For the low-emission scenario, these increases are on average 10% less. The results show that the mean annual daily baseflow is 157 mm/year (69% of discharge) for the present period and is shown to increase by percentages ranging from 30 to 90% between the 2020s and 2080s. Table 3 gives a summary of the simulated averages of the total annual precipitation, actual evapotranspiration, groundwater flow and the total streamflow using the semi-distributed model of WetSpa. It is obvious that there is an increase in the flow conditions and that greater changes are expected in the total streamflow. Table 3 Annual percentage change in simulations compared to the present period (1961–1990). Current annual A2 scenario: B2 scenario: Average: (1961–1990) 2010–2039 2040–2069 2070–2099 2010–2039 2040–2069 2070–2099 P = 1431 mm 12% 19% 43% 8% 16% 30% E = 1161 mm 5% 10% 25% 4% 8% 16% Rg = 159 mm 36% 50% 92% 21% 43% 71% Q = 268 mm 38% 60% 125% 24% 49% 88% P: precipitation, E: actual evapotranspiration, Rg: groundwater flow, Q: stream runoff.



722

(a) Discharge - A2a Scenario

0

50

100

150

200

250

300

350

Jan

Mar

May Ju

l

Sep

Nov

Dis

char

ge (m

m)

.

current2020s2050s2080s

(c) Baseflow - A2a Scenario

0

20

40

60

80

100

Jan

Mar

May Ju

l

Sep

Nov

Base

flow

(mm

) . current2020s2050s2080s

(e) Baseflow (% Discharge) - A2a Scenario

0

20

40

60

80

100

Jan

Mar

May Ju

l

Sep

Nov

Base

flow

(% d

isch

arge

) .


(b) Discharge - B2a Scenario

0

50

100

150

200

250

300

350

Jan

Mar

May Ju

l

Sep

Nov

Dis

char

ge (m

m).

.


(d) Baseflow - B2a Scenario

0

20

40

60

80

100

Jan

Mar

May Ju

l

Sep

Nov

Base

flow

(mm

) . current2020s2050s2080s

(f) Baseflow (% Discharge) - B2a Scenario

0

20

40

60

80

100

Jan

Mar

May Ju

l

Sep

Nov

Base

flow

(% d

isch

arge

) .


Fig. 7 Comparison of simulated changes in mean monthly river discharges and baseflows for both the high-emission (A2) and low-emission (B2) scenarios.

Figure 8 shows the spatial distribution of recharge obtained by running the fully-distributed WetSpa model. The results are presented for the current (1961–1990) and future periods (2020s, 2050s, 2080s) for both the high- and low-emission climate change scenarios. According to the results, the current average annual groundwater recharge in the catchment was estimated as 245 mm/year accounting to about 17% of the mean annual precipitation. Considering that there was no information available on groundwater conditions in the catchment, this recharge estimate cannot be fully verified. Generally, actual recharge estimates for most areas in Uganda remain unknown. However, a few available studies indicate estimates ranging from 100 to 220 mm/year (Taylor & Howard, 1996; DWD, 2005). The uncertainty in recharge estimates therefore justifies the need for more hydrogeological studies in Uganda in order to enable more reliable estimates for recharge.



723

(a)

(b)

Fig. 8 Simulated annual average recharge in mm due to climate change: (a) high-emission scenario (A2); and (b) low-emission scenario (B2).

For the high-emission scenario, the simulated average recharge in the 2020s is 360 mm/year, in the 2050s, 450 mm/year, and in the 2080s, 740 mm/year. This shows an increase in recharge predicted in the coming years, which corresponds to the predicted increase in precipitation as shown in Fig. 6. However, it is important to note that this study does not consider future changes in land use and effects due to future groundwater abstractions, which may play a significant role in the simulated recharge. The recharge varies across the catchment, as observed in Fig. 8, depending



724

on the land use and soil type. Recharge is more concentrated in the lower-lying areas along the river. This is explained by the fact that the higher areas are steep and, hence, experience more runoff. However, there is minimal spatial variation in recharge observed in the 2080s due to the high degree of wetness. CONCLUSIONS AND RECOMMENDATIONS This study investigated the effects of climate change on the groundwater system of the upper Ssezibwa catchment in Uganda. The hydrological model WetSpa, with a Nash-Sutcliffe prediction efficiency of 63%, was developed for the catchment and used to simulate the expected recharge and stream runoff under future climate change scenarios. The study does not include effects due to groundwater abstraction and land-use changes which may play a role in affecting future groundwater systems. Investigation of the historical data (1960–2006) in the catchment reveal evidence of present climate change based on increasing trends in temperature (approx. 2% per decade) and decreasing trends in discharge, especially from 1990 (approx. 50% per decade). No significant trends are observed in precipitation. This observation, however, cannot be relied on with great accuracy considering that only one station was used and that precipitation is highly variable in the catchment. These changes nevertheless justified the need for future climate predictions, which were based on SDSM-downscaled GCM predictors obtained from the HadCM3 climate model. The downscaled data show that both temperature and precipitation will increase in the future. Temperatures are expected to increase, especially in the dry season (May–September), by 1–3°C between the 2020s and the 2080s. Precipitation is also predicted to increase, with the greatest increases expected in the wet season (March–May; October–December) by over 30% in the 2020s, over 80% in the 2050s and over 100% in the 2080s. The WetSpa model shows that these climatic changes have a severe impact on the hydrology of the catchment. The simulations generally show an increase in recharge, baseflow and total flow in the future. The mean annual daily discharge is expected to increase by 40–100% from the current 1.47 m3/s in the coming 20–80 years. Corresponding increases in mean annual daily baseflows were found to range between 30 and 90% from the current baseflow of 0.87 m3/s, or 157 mm/year (69% of total runoff). Future simulations also indicate some seasonal variations in flows, whereby in the dry seasons almost all the river runoff will be entirely contributed by baseflow. It is simulated that the recharge will intensify in the future, this being caused by the predicted intensi-fication of rainfall. The current annual average recharge of 245 mm/year (17% of annual average precipitation) is expected to increase under intense greenhouse gas emissions (A2) by 47% in the 2020s, 84% in the 2050s and over 100% in the 2080s. However, the estimated current recharge could not be verified due to lack of groundwater information and discrepancies from previous studies. Thus, there is need for more hydrogeological studies in order to ensure more reliable recharge estimates. The results from this study generally suggest that, in the future, the region will experience increased annual flooding and rising groundwater levels, especially in the wet seasons, due to climate change. Given the limitations in data faced in this study, it is recommended to collect more ground-water data across the country to enable development of groundwater models that give more accurate predictions of the groundwater systems. More data for precipitation and temperature are also needed to improve both the hydrological and climate model projections. High-quality preci-pitation data in particular are needed to address the uncertainties normally contained in the downscaling process (Scibek & Allen, 2006). The findings from this study provide a basis for further research, especially in the downscaling of climate data and the need to perform sensitivity analysis of simulated hydrological changes. For the latter, it is necessary to establish the potential inconsistencies that may arise from calibrating the hydrological model with station data and running it using statistically downscaled gridded data. Hence, further research can investigate two alternative approaches: (1) where delta factors are applied on observed data and used as climate change inputs for the calibrated hydrological model; and (2) where the hydrological model is



725

calibrated with historical downscaled data and also run with downscaled future climate data. The results from these approaches can provide more reliable conclusions on future predictions. Acknowledgements This research was conducted within the framework of the InterUniversity Programme in Water Resources Engineering (IUPWARE) and was funded by the Flemish Inter-university Council, Belgium. The authors are grateful to the Meteorological Department and the Directorate of Water Development in Uganda, for providing the data for the research. We also thank Dr Antonis Koussis, Dr Richard Taylor, Dr Nicola Martin, Dr Anthony Duah and one anonymous reviewer for their comments and editing that greatly improved the final version of this paper. REFERENCES Arnell, N. W. (1999) Climate change and global water resources. Global Environ. Change 9, S31–S46. Blaney, H. F. & Criddle, W. D. (1950) Determining water requirements in irrigated area from climatological irrigation data.

US Dept Agric., Soil Conservation Service, Tech. Paper no. 96. Brooks, R. H. & Corey, A. T. (1966) Properties of porous media affecting fluid flow. J. Irrig. Drain. ASCE IR2, 61–88. Dibike, Y. B. & Coulibaly, P. (2004) Hydrological impact of climate change in the Saguenay watershed: comparison of

downscaling methods and hydrological models. J. Hydrol. 307, 145–163. Doherty, J. (2002) PEST: A Model-Independent Parameter Estimation User Guide. Water Numerical Computing, Brisbane

Australia. DWD (Directorate of Water Development) (2005) National Water Development Report for Uganda, 2005. UN World Water

Assessment Program (WWAP). DWD, Kampala, Uganda. Gleick, P. H. & Adams, D. B. (2000) The potential consequences of climate variability and change of the water resources of

United States. A special report for the US Global Research Program. US Geol. Survey. IPCC (Intergovernmental Panel on Climate Change) (2001) Climate Change 2001: The Scientific Basis. Contribution of

Working Group I to the Third Assessment Report of the IPCC (ed. by J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell & C. A. Johnson). Cambridge University Press, Cambridge, UK.

IPCC (Intergovernmental Panel on Climate Change) (2007) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the IPCC (ed. by S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor & H. L. Miller). Cambridge University Press, Cambridge, UK and New York, NY, USA (available at http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-spm.pdf).

Jónsdóttir, J. F., Jónsson, P. & Uvo, C. B. (2006) Trend analysis of Icelandic discharge, precipitation and temperature series. Nordic Hydrol. 37 (4-5), 365–376.

Joubert, A. M. & Hewitson, B. C. (1997) Simulating present and future climates of southern Africa using general circulation models. Progr. Phys. Geogr. 21, 51–78.

Jyrkama, M. I., Sykes, J. F. & Normani, S. D. (2007) The impact of climate change on spatially varying groundwater recharge in the Grand River watershed (Ontario). J. Hydrol. 338, 237–250.

Karl, T. R., Wang, W.-C., Schlesinger, M. E., Knight, R. W. & Portman, D. (1990) A method of relating general circulation model simulated climate to the observed local climate. Part I: Seasonal statistics. J. Climate 3, 1053–1079.

Kistler, R., Kalnay, E., Collins, W., Saha, S., White, G., Woollen, J., Chelliah, M., Ebisuzaki, W., Kanamitsu, M., Kousky, V., van den Dool, H., Jenne, R. & Fiorino, M. (2001) The NCEP/NCAR 50-year reanalysis. Bull. Am. Met. Soc. 82(2), 247–267.

Kulabako, N. R., Nalubega, M. & Thunvik, R. (2007) Study of the impact of land use and hydrogeological settings on the shallow groundwater quality in a peri-urban area of Kampala, Uganda. Sci. Total Environ. 381, 180–199.

Lee, K. K. & Risley, J. C. (2002) Estimates of ground-water recharge, base flow, and steam reach gains and losses in the Willamette River Basin, Oregon. US Geol. Survey Water Resour. Invest. Report 01–4215.

Liu, Y., De Smedt, F., Hoffmann, L. & Pfister, L. (2004) Assessing land use impact on flood processes in complex terrain by using GIS and modeling approach. Environ. Model. Assess. 9, 227–235.

Liu, Y., Batelaan, O., De Smedt, F., Huong, N. & Tam, V. T. (2005) Test of a distributed modeling approach to predict flood flows in the karst Suomuoi catchment, Vietnam. Environ. Geol. 48, 931–940.

Loaiciga, H. A. (2003) Climate change and groundwater. Ann. Assoc. Am. Geogr. 93(1), 30–41. Milly, P. C. D., Dunne, K. A. & Vecchia, A. V. (2005) Global pattern of trends in stream flow and water availability in a

changing climate. Nature 438, 347–350. Morris, B. L., Lawrence, A. R. L., Chilton, P. J. C., Adams, B., Calow, R. C. & Klinck, B. A. (2003) Groundwater and its

susceptibility to degradation: a global assessment of the problem and options for management. Early Warning and Assessment Report Series, RS.03-3. United Nations Environment Programme, Nairobi, Kenya.

Mitchell, J. (2005) Climate change and greenhouse effect briefing: a briefing from the Hadley Centre. Hadley Centre Metrological Dept, University of East Anglia, Norfolk, UK.

NEMA (National Environment Management Authority) (1997) State of Environment Report for Mukono District, 1997. NEMA, Kampala, Uganda.

O’Hare, G., Sweeney, J. & Wilby, R. (eds) (2005) Weather, Climate and Climate Change: Human Perspectives. Pearson Prentice-Hall Publishers, Harlow, UK.

Otim, M. (2005) The Nile basin water quality monitoring baseline report for Uganda. The Nile Basin Initiative—Transboundary Environment Action Project.



726

Phoon, S. Y., Shamseldin, A. Y. & Vairavamoorthy, K. (2004) Assessing impacts of climate change on Lake Victoria Basin, Africa. In: Proc. 30th WEDC Int. Conf. on Peoples Centred Approaches to Water and Environmental Sanitation (25–29 October, 2004, Vientiane, Lao PDR), 392–397.

Scibek, J. & Allen, D. M. (2006) Comparing modelled responses of two high-permeability unconfined aquifers to predicted climate change. Global Planet. Change 50, 50–62.

Taylor, R. G. & Howard, K. W. F. (1996) Groundwater recharge in the Victoria Nile basin of East Africa: support for the soil-moisture balance method using stable isotope and flow modelling studies, J. Hydrol. 180, 31–53.

Taylor, R. G. & Howard, K. W. F. (2000) A tectonic geomorphic model of the hydrogeology of deeply weathered crystalline rock: evidence from Uganda. Hydrogeol. J. 8, 279–294.

Tindimugaya, C. (2006) Overview of groundwater development in Uganda. Workshop for groundwater professionals in Uganda held August 2006, Kampala, Uganda [unpublished].

Wigley, T. M. L., Jones, P. D., Biffa, K. R & Smith, G. (1990) Obtaining subgrid scale information from coarse-resolution general circulation model output. J. Geophys. Res. 95, 1943–1953.

Wilby, R. L., Dawson, C. W. & Barrow, E. M. (2002) SDSM – a decision support tool for the assessment of regional climate change impacts. Environ. Model. Software 17, 145–157.

Willems, P. (2006) Trend Testing Tool. Hydraulics Laboratory, Katholieke Universiteit Leuven, Leuven, Belgium. http://www.kuleuven.be/hydr.

Willems, P. (2009) A time series tool to support the multi-criteria performance evaluation of rainfall–runoff models. Environ. Model. Software 24, 311–321.

Woldeamlak, S. T., Batelaan, O. & De Smedt, F. (2007) Effects of climate change on groundwater system in the Grote-Nete catchment Belgium. Hydrogeol. J. 15 (5), 891–901.

Xu, C. Y. (2000) Modeling the effects of climate change on water resources in central Sweden. Water Resour. Manage. 14, 177–189.

Revised paper received 23 January 2009; accepted 5 May 2009

estimating the effects of climate change on groundwater ... the effects of climate... · abstract...

Documents