VZJ Download to Citation Manager
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 23 August 2007
Published in Vadose Zone J 6:572-580 (2007)
DOI: 10.2136/vzj2006.0097
© 2007 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (3) Citing Articles via Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Aerts, J.
Right arrow Articles by de Vries, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Aerts, J.
Right arrow Articles by de Vries, A.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Aerts, J.
Right arrow Articles by de Vries, A.
Related Collections
Right arrow Global Change
Right arrow Recharge
Right arrow Runoff

SPECIAL SECTION: GROUNDWATER RESOURCES ASSESSMENT UNDER THE PRESSURES OF HUMANITY AND CLIMATE CHANGE

Robustness of Sand Storage Dams under Climate Change

Jeroen Aertsa,*, Ralph Lasagea, Wisse Beetsa, Hans de Moela, Gideon Mutisob, Sam Mutisob and Arjen de Vriesc

a Institute for Environmental Studies, Vrije Universiteit Amsterdam, De Boelelaan 1087, 1081HV, Amsterdam
b SASDL, P.O. Box 145333 Nairobi, Kenya
c ACACIA Institute, De Boelelaan 1085, 1081HV, Amsterdam
* Corresponding author (jeroen.aerts{at}ivm.falw.vu.nl).

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.


Received 8 July 2006.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Case Study: Kitui, Kenya
 Results and Discussion
 Conclusions
 REFERENCES
 
This study shows the robustness of subsurface storage using sand dams under long-term climate change for the Kitui District in Kenya. Climate change is predicted to enhance potential evaporation through an increase in average temperature of about 3°C. Even though average precipitation will also increase, approximately 13%, the net water availability is projected to decrease in the future, about 1 and 34% in the seasons November to March and April to October, respectively. This study shows that under current climate conditions, total storage in the 500 sand dams currently developed in Kitui captures only 1.8 and 3.8% from the total runoff generated during the November–March and April–October seasons, respectively. These numbers increase to 3 and 20% of total available water for the year 2100 for the November–March and April–October seasons, respectively. Hence, downstream water shortages can be expected under climate change in the April–October season. An additional water consumption scenario has been developed in which 1000 new sand dams are developed. In this case, the percentage storage by 1500 sand dams relative to the total available water increases to about 11 and 60% for the November–March and April–October seasons, respectively. In general, the variability in runoff is projected to increase under climate change, and the probability of years in which there is significant water shortage will increase from about once every 30 yr to once every 10 yr.

Abbreviations: EC, electronic conductivity • GCM, global circulation model • HadCM3, Hadley Climate Model 3 • IPCC, Intergovernmental Panel on Climate Change.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Case Study: Kitui, Kenya
 Results and Discussion
 Conclusions
 REFERENCES
 
It is widely accepted that climate change and climate variability will affect the hydrological cycle, resulting in more droughts and floods in some parts of the world (Arnell et al., 2001). Yet clear guidance on exactly how to respond to the challenge of climate change is lacking (Aerts and Droogers, 2004). This holds especially for many (semi-)arid regions located in developing countries in which rainfall is seasonal and highly erratic. For these areas, new developments in the area of groundwater management are seen as a key adaptation to climate change (IPCC, 2001).

In perennial rivers, most of the surface runoff occurs during short periods of heavy precipitation. Reservoirs can be developed to capture water, but they are vulnerable to loss of water through evaporation. The development of so-called sand dams, a special class of subsurface dams, addresses this issue (Nilsson, 1988; Ishida et al., 2003). Sand dams are constructed aboveground mostly as concrete structures between 10 and 100 m wide and 4 to 6 m high. Sand and soil particles transported during periods of high flow are allowed to deposit behind the dam (hence the term sand dam). The sand dam is constructed such that finer material will be washed downstream (Borst and Haas, 2006). When the space behind the dam is completely filled with sand (usually after 2–3 yr), water can be stored within the sand, thereby limiting losses from evaporation (Hellwig 1973a, 1973b). Water becomes available by pumping the water from the artificial sand aquifer that is created behind the dam.

To study the robustness of the sand dams under long-term climate change (the next 100 yr), research is needed that combines changes in potential water availability from sand dams with potential changes in water consumption. With these two trends, it is possible to evaluate the effectiveness of sand dams under climate change. However, little research has been conducted on the hydrogeological processes of sand dams. Hofkes and Visscher (1986) and Van Haveren (2004) described some general hydrological characteristics of sand dams, and more recently, Borst and Haas (2006) generated the first experimental results on the main hydrogeological fluxes of sand dams in Kenya. Furthermore, research on water consumption of communities that use water from sand dams are equally rare, except for studies conducted by, for example, Rempel et al. (2005) and Rhebergen and De Bruijn (2006).

The main goal of our research, therefore, is to set up a methodology to evaluate the effectiveness of sand dams for storing water under long term-climate change. The research is applied to a case study in the Kitui district in Kenya. Our objectives are


   Case Study: Kitui, Kenya
TOP
 ABSTRACT
 INTRODUCTION
 Case Study: Kitui, Kenya
Results and Discussion
 Conclusions
 REFERENCES
 
The Kitui District in Kenya is a semiarid region situated 150 km east of Nairobi (Fig. 1 ). The total land area is approximately 20,000 km2. The elevation of the district is between 400 and 1800 m. Hilly ridges (between 600 and 900 m in elevation) are separated by relatively flat areas. The district has approximately 555,000 inhabitants, of whom 58% lives beneath the poverty line of US$2 per day. The main economic activity is rainfed agriculture. Irrigated agriculture only takes place on small plots on the riverbanks. In the Kitui district, only 6% of the inhabitants have access to potable water. Water is the most essential development commodity in this area, and without artificial storage facilities, the major sources are ephemeral rivers. In times of water scarcity, people have to walk up to 20 km to collect water from major rivers, springs, and scoop holes.


Figure 1
View larger version (68K):
[in this window]
[in a new window]

  		FIG. 1. Location of the Kitui district in Kenya, with the gauging station used for calibrating the hydrological model the potential area for developing new sand dams.

 
The area is characterized by rainy periods that are highly erratic and unreliable. Rain usually falls during a few intensive storms (Nissen-Petersen, 1982; Borst and Haas, 2006). Only a minor part of the rain infiltrates in the soils on the slopes of the catchment, since they mainly consist of bare, low permeable clays and silts. Because the rainfall intensity is larger than the infiltration rate, a large part of the rain flows as overland flow. There are two rainy seasons, one from April to June (the "long rains") and one from October to December (the "short rains"). The terms short and long refer to the general intensity and duration of the precipitation events. On average, the precipitation in the Kitui District is approximately 800 mm year–1, but there are local differences in amount of precipitation due to topography and other influences. The potential evaporation is high, 1800 to 2000 mm year–1 (Zaal and Oostendorp, 2002).

Within the district of Kitui, about 500 sand dams have been developed over 10 yr to store water for the dry season. This makes it the area with the highest density of sand dams in the world. Since most precipitation falls in higher parts of the Kitui District, it is also the area where most of the agriculture potential is located and hence where sand dams are the most valuable. However, two major national parks lie further downstream that are partly drained by two rivers that have their origin in part in the higher areas of Kitui; The first is the Athi-Galana-Sabaki River, which flows through the Tsavo East National Park. The other major river is the Tana River, which also partly drains the northern parts of Kitui District. There is, however, no information on the water use or water requirements for these national parks, nor is information available on the domestic water use downstream of Kitui from the Tana and Athi rivers.

Sand dams differ from traditional dams by not storing water in upstream reservoirs but by storing water within the sand and gravel particles (≤600 µm) that accumulate against the dam and form an artificial aquifer (Fig. 2 ). Coarse gravel and sand can store and retain up to 35% of their total volume as water (Domenico and Schwartz, 1998). The subsurface reservoir is recharged through flash floods following rainstorms. When the reservoir is filled, surplus water passes the dam without infiltration. The stored water is captured for use by digging a scooping hole or constructing an ordinary well or tube well (Guiraud, 1989; Ishida et al., 2003). Figure 2 shows and example of two sand dams in a cascade. Sand dams are usually developed in valleys and regions with slopes that vary between 1 and 4% and with a favorable geology that generates sand as main sediment (Van Haveren, 2004).


Figure 2
View larger version (69K):
[in this window]
[in a new window]

  		FIG. 2. A schematic overview of two sand dams in a cascade and the main hydrogeological fluxes that determine the water balance of a sand dam catchment. See also Eq. [1] (from: Borst and Haas, 2006). (B, baseflow; Bs, lateral baseflow; E, evaporation from water in sand dam; ET, evaporation; Gr, groundwater flow; Lin, inflow from leakage through and around upstream dam; Lout, leakage from the riverbed to lower area; P, precipitation; Qin, river discharge inflow over upstream dam; Qout, river discharge outflow; R, direct runoff; Up, abstractable water.

 
The advantage of storing water in a sandy aquifer is that it prevents loss from evaporation (Hellwig, 1973a, 1973b). Hellwig (1973a) performed a number of experiments to relate the effects of grain size and water table depth to the rate of evaporation. Since water supply from the sand is not a limiting factor when the water table is at the sand surface, Hellwig (1973b) found that the grain size of sand does not affect the rate of evaporation when water level is at the surface. The rate of evaporation from a sand surface, however, is 8% lower than the evaporation from an open water surface, probably because of a higher storage of energy in a water body compared to a sand–water mixture, resulting in a larger temperature gradient between the evaporation surface and the air during the night (Hellwig, 1973b). When the water table drops below the sand surface, the rate of evaporation largely depends on the capillarity of the soil. Since the capillarity increases when the sand becomes finer, the rate of evaporation from fine sand will also be higher than that of coarse sand. Hellwig (1973b) found that for coarse sand (comparable to the sand generally found in Kitui), the rate of evaporation decreases to about 30% of the open water evaporation when the water table is at 30 cm below the sand surface, and to about 10% when the water table is at 60 cm below the sand surface. Because the sand dam aquifers usually measure 4 to 6 m high, most of the water that is stored within the sand is protected from evaporation.

Figure 2 also shows the main fluxes that allow for setting up a water balance for one individual sand dam. Borst and Haas (2006) have calculated these fluxes and estimated the potential extractable water. The extractable water U is defined as the amount of water that in theory could be harvested from the riverbed (‘the sand dam aquifer’) and riverbank plus the lateral baseflow Bs that adds to the amount in the riverbed during the season, minus evaporation from the riverbed and recharge to the basement aquifer. This is expressed by

Formula 1[1]
where Up is the abstractable water, Yr is the amount of extractable water from the riverbed, Yb is the amount of extractable water from riverbanks, Bs is the lateral baseflow, E is the evaporation, and Lout is the leakage from the riverbed to lower areas.

The calculations for the extractable water from both the riverbed Yr and riverbanks Yb can be obtained using

Formula 2[2]
where Yr/b is the amount of extractable water from either the riverbed or the riverbank (m3 season–1), L is the length of river segment (m), Wr is the average width of river (m), Dr is the average thickness of the riverbed (m), and Syr is the specific yield of the riverbed and riverbank fraction (%).

Using Eq. [1] and [2] with different values for the specific yields Syr (27% for riverbeds and 8% for riverbanks; Domenico and Schwartz, 1998), Borst and Haas (2006) found that the total potential amount of water that can be harvested in an average sand dam in Kitui is about 3100 m3 for the March–October season and about 5000 m3 for the November–February season. Hence, the maximum amount of water that can be harvested per year from an average sand dam is about 8100 m3. Alvarado (2006) differentiated these maximum estimations under the assumptions of having less lateral baseflow, Bs, and less extractable water from the riverbanks, Yb. Alvarado (2006) estimated an average of 2500 m3 for the April–October season and about 2800 m3 for the November–March season, resulting in a yearly maximum amount of water available of about 5300 m3.

The actual water consumption can be estimated using figures from Rhebergen and De Bruijn (2006) and Alvarado (2006). Water consumption can be classified into four categories: domestic, livestock, irrigation, and other uses (see also Nyong and Kanaroglou, 2001). Table 1 shows the average water consumption per category for both an average household and an average community using one sand dam by season and yearly. The table shows that only in the April–October season is the actual water consumption more than the available water within one sand dam.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Water consumption per season and year for both an average household and for a community that consumes water from one sand dam (Alvarado, 2006).

 
Climate Data
Historical data on daily precipitation and temperature for seven weather stations in the Kitui District have been derived from the Kenyan Meteorological Service for the period 1904 to 2004. Historical analysis of meteorological data shows that climate change is already an issue in the Kitui District. For example, Fig. 3 shows the frequency of yearly precipitation for seven meteorological stations in central Kitui for the period 1904 to 1954 together with the frequency of yearly precipitation for the period 1954 to 2004. The frequency of years with low annual precipitation has clearly increased in the period 1954 to 2004 compared with 1904 to 1954, and mean yearly precipitation decreased from 1000 mm in the first period to 800 mm in the second.


Figure 3
View larger version (12K):
[in this window]
[in a new window]

  		FIG. 3. Relative frequency of annual precipitation in Kitui over the period 1904–2004. The Figure shows annual precipitation has decreased in the second half of the past century as compared to the first half.

 
Climate change scenarios on monthly temperature and precipitation for the next century can be obtained from GCM (general circulation model) simulations. These models are forced by various greenhouse gas scenarios that describe different projected concentrations of greenhouse gas for the next 100 yr. The most commonly used emission scenario's are the A1, A2, B1, and B2 scenarios, ranging from relatively high to relatively low greenhouse gas emissions, that are reported in the Intergovernmental Panel on Climate Change's (IPCC) special report on emission scenarios (IPCC, 2001). For our study, the results from Hadley Climate Model 3 (HadCM3) under the A2 and B2 scenarios were selected for the period 1950 to 2100. The HadCM3 data are available on a 2.5° x 3.75° grid (~200 x 250 km2). We have used only one HadCm3 grid since the Kitui District falls completely into this cell.

To use the HadCM3 data for the Kitui case study, the data need to be downscaled to the appropriate spatial resolution and statistical properties. Two basic downscaling steps are needed (McGuffie and Henderson-Sellers, 1997; Aerts and Droogers, 2004). The first step is spatial downscaling of the coarse GCM data to a level that better reflects actual gradients and differences within the region. Based on experiences by Aerts and Droogers (2004) and Van Deursen and Kwadijk (1993), we use a resolution of 1 x 1 km2 in this study since this resolution allows for a visual inspection of the main streamflow pattern in the region. The spatial downscaling approach (Thiessen interpolation) is schematized in Fig. 4 .


Figure 4
View larger version (19K):
[in this window]
[in a new window]

  		FIG. 4. Schematic representation of spatial downscaling the course global climate data to a finer resolution that can be used for regional analyses.

 
The second step, statistical downscaling, is to transform the GCM output in such a way that the main statistical properties of historically observed data (1950–2004) match those of the transformed climate model output for the same period. The formula used for statistical downscaling GCM data is provided in Eq. [3], where the GCM data is not only corrected against the average observed climate but also for the observed variance (Aerts and Droogers, 2004).

Formula 3[3]
where a'GCM,i is the corrected climate parameter (total precipitation or average temperature) in month i, aGCM,i is the simulated climate parameter in month i, Formula 3GCM,i is the average simulated climate parameter in month i, {sigma}GCM,i is the standard deviation of the simulated climate parameter in month i, {sigma}obs,i is the standard deviation of the observed climate parameter in month i, and Formula 3obs,i is the average observed climate parameter in month i. The formula is constructed in such a way that both the average climate and the variability of simulated series after correction match the observation. For downscaling the GCM data, we used monthly precipitation and temperature data from HadCM3 and the measured weather station data—both over the period 1960 to 1990. For the same period, we used data from seven meteorological stations in the Kitui District.

According to the HadCM3 projections under the B2 and A2 scenarios, temperature and precipitation are expected to change quite significantly in the Kitui District. Temperature will rise roughly 2.8°C under the B2 scenario and about 4.1°C under the A2 scenario in the coming century (Fig. 5 ). Although historical data show a decreasing trend in precipitation over the last 100 yr, the HadCm3 projections predict a weak increase in future precipitation of about 6 to 8% over the period 2050 to 2100. The variability, however, is projected to further increase as similarly shown in the historical precipitation data.


Figure 5
View larger version (51K):
[in this window]
[in a new window]

  		FIG. 5. Projected monthly temperature and precipitation data under scenarios B2 and A2 (IPCC, 2001) generated by the Hadley Climate Model 3 (HadCM3) for Kitui. The graphs represent moving averages (mov. avg.) of 12 mo.

 
Hydrological Model
The STREAM model is a spatial water balance model that calculates runoff on the basis of temperature and precipitation data and a number of land surface characteristics (Van Deursen and Kwadijk, 1993; Aerts et al., 1999). The STREAM model for the Kitui District uses spatial data at a spatial resolution of 1 x 1 km2 and at a monthly time step.

The water balance is calculated for each grid-cell using a direct runoff, soil water, and groundwater component according to a number of parameters (Aerts et al., 1999). Total runoff, T, is calculated as

Formula 4[4]
where R is direct runoff, M is snow melt, and B is the base flow origination from groundwater (all mm mo–1).

The direct runoff R is calculated from the soil water balance, S, using a separation coefficient sc:

Formula 5[5]
The remaining amount of water from the soil water balance is redirected to the groundwater (TG, to groundwater flwo), using

Formula 6[6]
The base flow is calculated from the amount of groundwater (GW) stored using a recession coefficient rc:

Formula 7[7]
The soil water balance and actual evaporation are calculated for each month using the equations from Thornthwaite and Mather (1957). Actual evaporation, EA0', is estimated from adjusted reference evaporation, ET0, using a crop factor kc and a reduction coefficient Fred that acts as calibration factor:

Formula 7[8]
Reference evaporation is calculated from temperature, using the formulas from Thornthwaite (1948), and FAO factors are used for adjusting the reference evaporation to different land-cover types using crop factors (Doorenbos and Pruitt, 1975). Land-cover classes are taken from the Global Land Cover Characteristics database Version 1.2, produced by the International Geosphere Biosphere Program. This data set is based on NOAA AVHRR satellite observations from April 1992 to March 1993, which were classified to land-cover characteristics by Belward et al. (1999). Parameters for the maximum soil water holding capacity are taken from a global data set compiled by the USDA (available from http://www.soils.usda.gov/use/worldsoils/index.html) with a resolution of 2 arc minutes (about 3.5 x 3.5 km2). These parameters are processed according to the model script presented in Fig. 6 .


Figure 6
View larger version (37K):
[in this window]
[in a new window]

  		FIG. 6. Part of the STREAM model script showing the calculation of both the potential evapotranspiration (ET) and actual (AE) evapotranspiration and the soil water components. (Thornthwaite and Mather, 1957.)

 
The digital elevation model is derived from the Shuttle Radar Topography Mission data set (90 x 90 m2; http://srtm.usgs.gov/) and has been processed to a resolution of approximately 1 x 1 km2. The digital elevation model is used for streamflow routing.

The STREAM model has been successfully applied in various forms to climate and hydrology studies in a number of river basins with similar size and characteristics as the Kitui District (Van Deursen and Kwadijk, 1993; Aerts et al., 1999; Middelkoop et al., 2001; Winsemius et al., 2006). These studies have confirmed that a monthly time step is sufficient for detecting decadal, inter-annual and seasonal changes in the hydrological cycle, such as those caused by water consumption and climatic change.

The STREAM model was calibrated on the basis of the downscaled HadCm3 data for mean monthly discharge values over the period 1957 to 1979 since for this period, mean observed discharge data for the outflow of the Upper Athi River basin (which drains the Kitui District) has been used (LBI, 1983). The calibration of the model involved the adjustment of a reduction factor that tunes the reference evaporation (see Eq. [8]), a coefficient that determines the separation between groundwater and runoff (Eq. [5]), and a recession coefficient that determines the delay of the groundwater flow (Eq. [7]). The calibration also involved the match to observed total annual runoff, as well as seasonal patterns using the efficiency coefficient from Nash and Sutcliffe (1970). The model efficiency coefficient after calibration of R2 = 0.73 indicates that the model is capable of reasonably estimating mean monthly runoff. The correlation coefficient (r2) between observed and simulated monthly runoff is 0.91. Note, however, that although calibration results are good, we were not able to perform a validation since only mean monthly runoff data were available. The calibration results are depicted in Fig. 7 .


Figure 7
View larger version (14K):
[in this window]
[in a new window]

  		FIG. 7. Calibration results for Kitui District showing the mean monthly measured runoff and the mean simulated runoff.

 

   Results and Discussion
TOP
 ABSTRACT
 INTRODUCTION
 Case Study: Kitui, Kenya
 Results and Discussion
Conclusions
 REFERENCES
 
To determine the projected future trend in water availability for the Kitui District, the downscaled HadCm3 results under the A2 scenario have been used as input to the calibrated STREAM model. From here, water availability is defined as the surface runoff simulated by STREAM. As water consumption is dependent on the two distinct rainy seasons, water availability has been determined per season for the period 1950 to 2100. Figure 8 shows the trend in monthly runoff distribution over three time slices (1960–1990, 2020–2050, and 2070–2100). Data analysis shows that the mean runoff decreases under climate change with about 1 and 34% in the seasons November to March and April to October, respectively (Table 2). However, from Fig. 8 it can be derived that in both seasons both extreme low and high runoff figures increase as well, especially in the April–October season. For example, the number of April–October seasons with runoff figures lower than 50 mm increases significantly over time. This confirms the projected trend of increasing hydrological extremes under climate change such as described by the IPCC (2001).


Figure 8
View larger version (16K):
[in this window]
[in a new window]

  		FIG. 8. Trend in monthly runoff distribution (in mm) over three time slices (1960–1990; 2020–2050 and 2070–2100) for the Apr.–Oct. season (left) and the Nov.–Mar. season (right).

 

View this table:
[in this window]
[in a new window]

  		TABLE 2. Change in hydrological parameters for three time slices: 1960–1990; 2020–2050; 2070–2100, with potential evapotranspiration (ET), actual evapotranspiration (AE), precipitation (PRE), and effective precipitation (PEFF = PRE – ET).

 
Using both current climate conditions and the maximum amount of yearly harvested water of 8100 m3, Borst and Haas (2006) estimated the percentage water stored by the 500 sand dams relative to the total runoff as 2.3% for the April–October season and 2.5% for the November–March season. Note, however, that Borst and Haas (2006) only observed a small part of the Kitui District and estimated total runoff using a runoff coefficient. We can now calculate the percentage of water stored by the 500 sand dams using the total water available simulated by STREAM over the period 1960 to 1990. It appears that the percentage storage per season relative to the total runoff per season amounts to 3.8% for the April–October season and 1.8% for the November–March season. Hence, the numbers are of similar magnitude. However, Borst and Haas (2006) estimated the total runoff for each season to be almost equal. In our study, the average total runoff from November to March is higher than the total runoff from April to October (Table 2).

Furthermore, the same calculations can be made over future time slices using two scenarios: 500 and 1500 dams. Figure 9 shows the trend in storage of the 500 dams as a percentage of the available water over the period 1950 to 2100 (dashed line), indicating how much water (in %) both 500 dam and 1500 dams take from the total available water. One hundred percent means that there is just enough water to fill all the dams. The vertical bars in Fig. 9 represent the percentages for both 500 and 1500 dams. For the November–March season, the total storage of 500 dams remains about 2% of the total available water; although water availability becomes lower under climate change, the percentage storage relative to the total available water remains negligible. A remarkable difference can be seen when studying the same calculations for the April–October season. The storage of 500 dams as a percentage of total available water increases from about 3% in the 1950s to about 20% in the year 2100. Hence, it can be derived that with the existing storage capacity of 500 sand dams, downstream water shortages can be expected under climate change in the April–October season.


Figure 9
View larger version (37K):
[in this window]
[in a new window]

  		FIG. 9. Trend in storage of 500 and 1500 sand dams as a percentage of the available water over the period 1950–2100 for the November–March season (upper graph) and the April–October season (lower graph).

 
Furthermore, water consumption is expected to rise in the future through population growth, increase in irrigation, and increases in new water-consuming activities (e.g., brick making). We therefore assume a scenario in which 1000 new dams will be developed over the next 100 yr to meet this demand. We further assume that the total available water in one sand dam is fully consumed, which is a fair assumption considering the consumption figures from Table 1 and the available water in a sand dam discussed in the previous section. We have not assumed a specific spatial distribution of the additional 1000 sand dams in the district, since alternative spatial distribution are not relevant for the timescales used in this study (~100 yr) (for more details on spatial distributions, see van Loon and Droogers, 2006).

If we analyze a scenario of 1500 dams for the November–March season, the percentage storage of the total available water increases to about 11% over the period 1950 to 2100—this will imply some influence on downstream water availability. For the April–October season, changes are significant; the percentage water stored by 1500 dams increases to about 60% of the total available water projected for the year 2100.

Because there is often not enough water projected in the sand dams to cover the current average community water consumption per sand dam, and because the future projections indicate the April–October season may generate most of the problems with water shortage, we now focus on this season. Using current consumption numbers, the total water consumption from 500 sand dams over April to October presently amounts to about 1,300,000 m3. Supposing that 1000 sand dams are needed to fulfill extra water demands, then the total consumption is estimated to about 3,800,000 m3. This trend, together with the trend in water availability for the April–October season, is depicted in Fig. 10 . The figure only indicates very dry April–October seasons with water availability numbers <10,000,000 m3. It shows that extreme water deficiency occurs currently about once every 30 yr. Under the assumed consumption scenario and the water availability under climate change, April–October seasons with extreme water deficiency will increase to once in every 10 yr. Sand dams will become more important during this season since the potential for water rainfed agriculture will decrease. This can be derived from Table 2, where a lower future number for both effective precipitation and actual evaportranspiration indicates less water will be available for rainfed crop growth.


Figure 10
View larger version (12K):
[in this window]
[in a new window]

  		FIG. 10. The trend in water consumption over the period 1950–2100. All symbols below this line indicate extreme dry April–October seasons, where water demand cannot be met. Only extreme dry April–October seasons are depicted with seasonal water availability <10,000,000 m3.

 

   Conclusions
TOP
 ABSTRACT
 INTRODUCTION
 Case Study: Kitui, Kenya
 Results and Discussion
 Conclusions
REFERENCES
 
The objective of this study was to analyze the robustness of subsurface storage using sand dams under long-term climate change (next 100 yr). A water balance model—STREAM—was developed to simulate the water availability under climate change. Climate change scenarios were derived from the HadCm3 model, which provides monthly precipitation and temperature data for the past 50 and next 100 yr under different greenhouse gas scenarios. The model was applied to the Kitui District in Kenya.

Under current climate conditions, this study shows that total storage in the 500 sand dams that are currently developed in Kitui only captures 3.8 and 1.8% from the total runoff generated during the April–October and November–March seasons, respectively. These numbers increase to 20% (April–October) and 3% (November–March) for the year 2100. Hence, it can be derived that with the existing number of 500 sand dams, downstream water shortages can be expected under climate change during the April–October season. An additional water consumption scenario has been developed in which 1000 new sand dams are developed. In this case, the percentage storage by 1500 sand dams relative to the total available water increases to about 11% and 60% for the November–March and April–October seasons, respectively.

In general, the variability in runoff is projected to increase under climate change and the probability of years in which there is significant water shortage will increase from currently about once every 30 yr to once every 10 yr in the year 2100.

More research is needed regarding the physical processes that pertain to the functioning of sand dams:

Additional research is also needed to derive the mechanisms between water consumption and water availability. In this study, we assumed that there is a linear relationship between water use and water availability, which is currently the case (Rhebergen and De Bruijn, 2006). However, extreme droughts or wet situations could change the water consumption patterns, and these should be taken into account in new water consumption scenarios.


   ACKNOWLEDGMENTS
 
S. Mutiso, P. van Dongen, E. Odada, Calistus Wachana and C. Mutiso. Meteorological data were provided by the Meteorological Department of Kenya, Nairobi.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Case Study: Kitui, Kenya
 Results and Discussion
 Conclusions
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
T. R. Green, M. Taniguchi, and H. Kooi
Potential Impacts of Climate Change and Human Activity on Subsurface Water Resources
Vadose Zone J., August 23, 2007; 6(3): 531 - 532.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (3) Citing Articles via Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Aerts, J.
Right arrow Articles by de Vries, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Aerts, J.
Right arrow Articles by de Vries, A.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Aerts, J.
Right arrow Articles by de Vries, A.
Related Collections
Right arrow Global Change
Right arrow Recharge
Right arrow Runoff

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome