THE EFFECT OF LENGTH OF TIME PERIOD CONSIDERED IN ESTIMATING GROUNDWATER RECHARGE WITH A SOIL WATER BUDGET MODEL

The rate of replenishment of water table (or rate of groundwater recharge) is a key issue one needs to know in developinggroundwater resources. The soil water budget method is often used to estimate this rate of recharge. The need to use daily climatic data (rather than weekly or monthly data) in the soil water budget has been shown by Howard & Lloyd? However, no recommendations exist on the minimum duration necessary for precise soil water budgeting. This study attempts to find a suitable answer to this question, choosing Angunakolapeiessa in the dry zone of Sri Lanka and Silsoe in UK as study locations. The results of the study show that it is good practice to use as many years of daily water balance data as possible. If this is not possible, climatic data for a t least 3 years (consisting of a wet year, a dry year and an average year) must be used in order to achieve realistic estimates ofrecharge.

The ratio of ETa/ETp when soil moisture deficit is greater than root constant Field capacity of soil (%) Interception (rainfall) storage capacity (mdday) Preferential flow coefficient Threshold of daily rainfall above which preferential flow occurs (mmday) Preferential flow (i.e., flow of infiltrating rainwater through cracks and fissures, bypassing the soil matrix) Permanent wilting point of soil (%) Rainfall (mdday or m d y ) RC = Root constant (% ofAWC) = soil moisture deficit at which water stress begins to effect ETa ROC = Runoff coefficient Rot = Threshold of daily rainfall above which runoff occurs (mmlday) RZD = Root zone depth (mm) SMD = Soil moisture deficit (mm) The r a k , I " recharge is defined as the rate of replenishment of the water table. A knowledge of the rate of recharge must be available in developing a groundwater source for domestic, industrial or agricultural purposes, as this parameter determines the rate of safe yield that can be abstracted from the groundwater resem0ir.l It is very difficult to determine the rate of recharge to the water table accurately. The best hope is to arrive at a reasonable estimate. There are a few commonly used methods of estimating re~harge.~",~ They are (a) the use of lysimeters (b) soil water balance models (c) water table fluctuation method (dl catchment water balance method (el numerical modelling of the unsaturated zone (f) zero flux plane method (g) Darcy method (h) Tritium method and (i) Chloride method.

1
Of the above methods, the soil water balance method is a simple and an easy to use one in most climatic conditions and quite often is the only available method for a particular climatic ~ondition.~ In this method, a volume balance for the water entering and leaving the root zone and change in soil moisture storage is carried out and recharge (Re) is estimated as; Here P is precipitation, 1 is interception of rainfall by vegetation, RO is run off, PF is preferential flow, ETa is actual evapotranspiration and A S is change in soil moisture storage. It is important to note that the first term within brackets is the matrix flow (MF) and the second term within brackets is the preferential flow (PF). Now, the estimation of actual evapotranspiration is affected by matrix flow, which in turn is affected by the amount of preferential flow. Therefore, estimates of recharge (which are affected by estimates of actual evapotranspiration), are affected as a result of preferential flow (i.e., for estimates of recharge to be affected by preferential flow, it is not necessary for preferential flow paths to be effective for deeper depths, but depths just around root zone are sufficient).
If the balance is carried out annually (especially from the end of the rainy season to the same time the following year), the change in soil moisture storage is negligible (as moisture content at both times will be at field capacity). Therefore, equation 1 can be reduced' to; (2) Usually, recharge estimated by equation (2) is computed for a few years (normally one or two years) with a daily time step.6 However, since rainfall varies significantly from year to year (especially in a country like Sri Lanka), the resulting estimates of annual recharge also vary annually. This paper studies the effect of the length of \ time period considered on the reliability of estimating recharge with a soil water budget model.

METHODS AND MATERIALS
, The methodology adopted in general was to estimate recharge for different -time periods (for different locations) and compare the estimates. The specific steps of the methodology adopted were as follows: (a) Select suitable study locations.
(b) Collect relevant data (i.e., daily rainfall and pan evaporation for a number of years, information on rainfall interception and runoff). Bedfordshire in England, UK and Angunakolapelessa in the southern dry zone of Sri Lanka (Fig.1). These two locations were chosen because of..the availability of long term rainfall and evaporation data that were needed for the study and also because of the availability of other required data such as field capacity and permanent wilting point of the root zone soil. Table 1 summarises the climatic, vegetative and top soil details at the study locations.

Determination of field capacity, wilting point and other parameters: A summary of
climatic data and soil properties at each location is shown in Table 2. Details of climatic data and details of experiments carried out to determine the soil properties in Table 2 are given by de S i l~a .~ The field capacity and permanent wilting point were measured using a pressure plate apparatus as described by de S i l~a .~    Here, the value of p is equal t o the root constant. Values of pan evaporation were converted to values of evapotranspiration using pan co-efficient^.^ Table 3, below, gives the values of Isc, Rot, ROC, PFt and PFc used for the two locations. These values were obtained from de Silva.'  Table 4 and Table 6 give the estimates of recharge for Angunakolapelessa and Silsoe respectively, if the budgeting period is considered as one year. (Some studies have suggested that ETa is normally higher than what is estimated in Column 6 of Table  4 for Angunakolapelessa). However, in this study, ETa was estimated as shown in Fig. 2 Table 3 are reasonably correct for the two areas of study. Details of how these model parameters were decided are given by de S i l~a .~ As expected, the soil water budget model correctly predicts that a t Angunakolapelessa, most components of the hydrological cycle (interception, runoff, evapotranspiration and recharge) are significant (Table 4), but at Silsoe, only evapotranspiration and recharge are significant ( Table 6). The reasons for zero runoff at Silsoe is that the soil at the study location is sandy (Table 1) and also the precipitation is low intensity spread over 9 months of the year (except during the summer months). However, at hgunakolapelessa, high intensity rains of shorter duration on hard soils lead to significant amounts of runoff. Average recharge for the years considered ' Using the soil water budget model shown in Fig. 2, recharge was estimated for both study locations for different lengths of time periods [ie forAngunakolapelessa the durations were one year, 3 years, 5 years, 7 years and so on and for Silsoe the lengths of time periods were one year, 4 years, 8 years and so on. Fig. 3(a) shows the estimates of recharge for Angunakolapelessa for time periods of one year and Fig.  3(b) shows the estimates of recharge for time periods of 3 years (i-e., for 1976,1977 & 1978 and 1977,1978 & 1979 and so on). Fig. 3(c), 3(d) and 3(e) show the estimates of recharge if the time period is considered as 5, 7 and 9 years respectively. From Fig. 3(a) the estimate of recharge could vary from about 20 to 150 mmfy, based on a time period of one year. When the time period is increased, this variation is reduced as shown in Fig. 3 (b), (c), (dl and (e) and as shown in Fig. 5. Similarly, Fig. 4 Table 4 shows that the lowest recharge (20 m d y j was in 1989 and the highest recharge (150 mmly) was in 1977 at Angunakolapelessa. Table 6 shows a similar result for Silsoe; the lowest recharge of 17 mmfy was in 1964 and the highest recharge of 241 mmfy was in 1979. Fig. 5 and Fig. 6 show the ranges of recharge obtained if different time periods are considered at Angunakolapelessa and Silsoe. It is seen fmm Fig. 5 and Fig. 6 that it is not possible to decide on a suitable (optimum) time duration (in a soil water budget) for a particular location.  (1989), wettest (1982) and an average year (1987) Year P Average recharge for the years considered 72 Table 5 shows the estimates of recharge for the wettest (1982), driest (1989) and an average year of rain (1987) during the time period considered (i.e., from 1976 -1991) for Angunakolapelessa. At the bottom of Table 5, the average recharge value obtained by considering only these 3 years (ie the wettest, driest and an average year) are shown, which is remarkably close to the average of the estimate of recharge in Table 4. Similarly, Table 6 shows the estimates of recharge for Silsoe with the time duration considered as one year (with an average recharge of 116 mmly) and Table 7 shows the estimates of recharge for the wettest (19641, driest (1979) and an average year (1980) for Silsoe (with an average recharge of 124 mmly). As seen from these two tables (ie Tables 6 & 7) the two average estimates of recharge obtained by considering a full 25 years (in Table 6) and by just considering 3 years (in Table 7) are similar.  From the above discussion, it is clear that one is not able to conclude that to estimate recharge at a particular location with a soil water budget, consideration.of a certain number of years of climatic data is sufficient. As shown, climatic data ofgas many years as possible need to be considered if reasonably true values are to be obtained. Considering the results of Tables 5 and 7 (and comparing the results of Table 4 and Table 6 respectively), it is also reasonable to conclude that in situations where a number of years of climatic data cannot be considered, at least a wet, dry and an average year need to be considered in a soil water budget to estimate recharge.   Average recharge for the years considered 124

CONCLUSION
In estimating recharge with a soil water budget in a particular location, climatic data for as many years as possible should be used for the location considered. As has been done in many studies, consideringjust one year's data (or even 2-3 years of data) will certainly not be sufficient and will lead to recharge estimates which can vary significantly. However, if this is not possible for some reason (for example, limited data availability or non availability of suitable computing capabilities), then at least a wet, dry and an average (normal) year (with respect to mean annual rainfall) must be considered in estimating recharge with the soil water budget method. The final value of recharge may be taken as the average of the recharge estimates for these wet, dry and the normal year.