STUDY OF ARTIFICIAL GROUNDWATER RECHARGE FROM A POND IN A SMALL WATERSHED

Abstract

Artificial groundwater recharge (AGR) by conservation of surface runoffs on watershed basis, as a measure to restore back the depleted groundwater level and to augment the groundwater resource in phreatic aquifer, is promoted by both the central and state governments in India. It is as one of the key strategies, for sustainable groundwater management, particularly in the arid and semi-arid regions. A number of AGR schemes are practiced in India; in which, practice of groundwater recharge by pond is very common due to its constructional simplicity and low operational and maintenance cost. For sustainable groundwater management and for economic design of well based on the augmented water from such recharge schemes, one has to know the rate of recharge and its responses to the underneath aquifer. Each recharge pond has a specific catchment area from where it receives water. The runoff that generates from the rainfall over the pond's catchment, stores in the pond and then infiltrates below for recharge to the aquifer at a rate faster than the normal recharge. The rate of recharge depends on the difference of heads in the pond and the underneath groundwater table in addition to the subsurface soil properties and the saturated hydraulic conductivity. The head of water in the pond is governed by the: (i) surface runofffrom its catchment, (ii) direct rainfall over its surface, (iii) evaporation from its water surface, and (v) excess outflow runoff from its storage. In the present investigations, each of the hydrologic components involved in the water balance of a pond has been studied separately to derive suitable process level models in accordance to the data usually monitored in the field, and thereafter, the derived process level models are integrated in the water balance equation of the pond to arrive at the required recharge estimation models. The recharge component that constitutes two aspects; the potential recharge and the actual recharge, has been studied using the Green-Ampt (GA) infiltration model for deriving the process based potential recharge model, and the Hantush's approximate analytical solution for deriving the process based actual recharge model. Approximating the logarithmic term, ln(l +Lf/H +v|/f) of the GA model [where Lf = length of the advancement of the wetting front; H = depth of ponding; and v|/f = suction head) by the segmental second order polynomials; the approximate expressions for Lf are derived satisfying all ranges of [Lf/(H+\(/f )]> 0. Five segmental equations are found to fit to ln[l +Lf/(H +\|/f )J for [Lf/(H +\|/f)] >0 within the error bound of ± 1%. Unlike estimation of Lf by the trial and error method as required in the GA model, the proposed expressions estimate Lfexplicitly. The characteristic behaviors of the derived Lfexpressions are studied for four textural classes of soils namely; sand, loam, clay loam and sandy clay, considering depth of water to be constant, and are found have similar responses as described by the GA model. The proposed model for/-/is thereafter used to develop the time varying process based potential recharge estimation model. The responses of the potential recharge model are studied for all the four soil texture groups. The results showed similar behaviors as that of the GA model. The equation to estimate the time for the wetting front to the reach water table is also derived. Using the Hantush's approximate analytical equation that has been derived for predicting the rise of groundwater level due to constant recharge through rectangular basin, the process based actual recharge rate estimation model for time varying depth of water and water spread area is developed. The characteristic behavior of the actual recharge estimation model is also studied. By linking the normalized antecedent precipitation index (NAPI) with the antecedent soil moisture content (AMC) and AMC to the losses of precipitation, a simple rainfall-runoff model for predicting runoff yields is derived using the basic water balance equation. The ii model is based on the concept of normalized antecedent precipitation index proposed by Heggen (2001). The model has three watershed specific parameters, which can easily be estimated from historical data of rainfall-runoff events. The model requires single input; rainfall and one rainfall dependant variable 'NAPE. The performance of the proposed model is compared with the Soil Conservation Service-Curve Number (SCS-CN) model. The parametric relationships between parameters of the proposed model and CN of the SCS-CN model are also studied. The proposed model is tested with the field data collected from three small watersheds located in the semi-arid region in Rajasthan, India. The results exhibited a superior match by the proposed model than the SCS-CN model. By evaluating the performances of four commonly used evaporation estimate models, namely; Bowen ratio energy balance (BREB), Mass transfer (MT), Priestley-Taylor(PT) and Pan evaporation (PE), based on the experimental field data, the most effective and reliable model for estimating the evaporation rate for the semi-arid region in Rajasthan is identified. The performances of the BREB, the PT and the PE models are found nearly complementary to each other while the MT method is found to deviate. In this study, the BREB model is used for the computation of evaporation rate, as the data required by the BREB model were available from the field investigation. By extending the process based models derived for rainfall-runoff, evaporation and the recharge estimation in the water balance equation of a trapezoidal pond, the integrated models for computing the time varying depth of water, the potential and actual groundwater rechargehave been developed. The models are developed for two different sets of condition: first, till the wetting front touches the water table, and the second one, as the water table rises due to subsequent recharges after the pond is hydraulically connected to the aquifer. The performances of the integrated models have also been tested using the data collected from the experimental trapezoidal pond located in the semi-arid region in Rajasthan. The comparison of the observed and the simulated depths of water are found to have a close in match. The correlation coefficients of the statistical parameters have also shown a good agreement between the observed and simulated results. The responses of the integrated process based models in estimation of the potential and the actual recharge rates are found most promising. These process based models can be extended to other areas for quantifying the recharge component for similar or other types of recharge schemes. For evaluating the performances of two data series, the statistical parameters namely; coefficient of determination, index of agreement, percentage relative error, standard error of estimate, and relative bias are chosen as the guiding factors.