Résumé
This study focuses on 11,000 km(2) of the Musi River sub-basin which is one of the main tributaries of the Krishna River, located in Andhra Pradesh (South India). The basin has a semi-arid climate with precipitation occurring during the rainy season, from June to October and with a total average of 710 mm yr(-1). During the past, the watershed development has led to significant changes in land use. Now, nearly 70% of the basin is cultivated of which 45% is irrigated. Around 60% of the water for irrigation is supplied by groundwater extraction. The number of bores in use has increased ten times from 1991 to 2001 and should currently exceeds 45,000 (on average, 1 active bore for 4.5 ha irrigated). The Musi Medium Irrigation project covers 12,500 ha and the Nagarjuna Sagar Left Canal supplies water for 43,000 ha downstream of the Musi sub-basin. Wastewater of mixed domestic and industrial origin is utilized to irrigate approximately > 10,000 ha of paddy rice along the Musi River in peri-urban and rural Hyderabad. More than 1160 artificial percolation tanks had been built on the Musi catchment drainage network.
The preliminary analysis of more than 60 groundwater level time series widespread all over the sub-basin (from 1989 to 2004) shows a general long term depletion trend of the water table while no significant rainfall deficit was observed over the same period. The average rate of depletion is estimated as 0.18 m yr(-1) with maxima in some areas of up to 0.40 m yr(-1) (Figure 1). This situation can be explained mainly by groundwater exploitation, a consequence of the watershed development for agriculture. The Musi sub-basin is mainly covered by Archaean granites with Deccan Traps at the Eastern edge. As in a typical hard rock aquifer region, the yield of the bores decreases with depth due to the reduction of the fracture density. Hence the risk of water scarcity in case of a drought year is exacerbated. In order to assess aquifer renewable reserves and help groundwater management authorities, a fully distributed physical model of the aquifer has been calibrated and validated for a transient state experienced during 1989-2004 by using MODFLOW. The key variables such as aquifer storativity and transmissivity were determined by inverse fitting of simulated and observed groundwater levels.
[GRAPHICS]
Mean annual simulated recharge is 1176 Mm(3) (17% of total rainfall) while annual pumping is estimated at 1235 Mm(3). Simulated base flow is 23 Mm(3) while river leakage is less than 1 Mm(3). Among the total simulated annual recharge, groundwater irrigation return flow to the aquifer can be estimated at 370 Mm(3) (31%) and artificial recharge at 124 Mm(3) (11%). Natural recharge from rainfall accounts for 652 Mm(3) (55%). It is close to 9% of the total annual rainfall. The sustainable groundwater withdrawal yield over the period is around 1220 Mm(3) for the total basin. A deficit of 124 Mm(3) for the long term groundwater balance (16 years) justifies the observed depletion trend of the water-table of -0.18 m yr(-1).