Estimation of net groundwater use in irrigated river basins using geo-information techniques : a case study in Rechna Doab, Pakistan

Keywords: remote sensing, GIS, water balance, groundwater, net groundwater use, recharge, water management, Rechna Doab, Pakistan.Over-exploitation of groundwater resources threatens the future of irrigated agriculture, especially in the arid and semi-arid regions of the world. In order to reverse this trend and to ensure future food security, the achievement of sustainable groundwater use has become a global issue. To address the challenges of aquifer mining and sustainable management, a quantitative analysis of recharge and groundwater use is essential. Spatio-temporally distributed information on <em>net groundwater use</em> - i.e. the difference between tubewell withdrawals for irrigation and net recharge - is often unknown at the river basin scale. Conventionally, groundwater managers and policy makers look at either groundwater withdrawal by tubewells or phreatic surface fluctuations (if available). However, these methods are ineffective for describing the vertical water fluxes occurring between the unsaturated and saturated zones. Even in detailed groundwater modelling, these vertical water fluxes are difficult to assess.This endeavour aspires to develop a methodology for computing the various water balance components of the unsaturated zone by using geo-information techniques. These water balance components are then used to compute the net groundwater use. With this approach, groundwater recharge will not be quantified explicitly, but is a part of net groundwater use. Records of routine climatic data, canal discharges at major offtakes, phreatic surface depth, and coarse information on soil textural properties are required as input data. The Rechna Doab region (approximately 2.97 million ha), located in the Indus basin irrigation system of Pakistan, has been used for this case study.In order to better understand the interaction and dynamics of sub-soil water fluxes, field studies were conducted in rice-wheat and cotton-wheat areas representing shallow (2 m), as well as deep (10 m), phreatic surface conditions. A detailed physically-based transient agro-hydrological model (SWAP) has been used to compute sub-soil water fluxes including recharge, and capillary rise. The SWAP model was calibrated using <em>in situ</em> measurements of soil moisture content and actual evapotranspiration. The results of the field modelling were used to develop and test a new, simple method for computing soil moisture storage changes in the unsaturated <em></em> zone, using root zone soil moisture content and depth to the phreatic surface. This method can also be applied in combination with Remote sensing and GIS data to compute soil moisture storage changes across vast areas.Regional scale actual evapotranspiration and soil moisture maps were derived using the remote sensing algorithm termed SEBAL (Surface Energy Balance Algorithm for Land). New geomatic approaches were developed and tested that estimate the disaggregated canal water distribution in an irrigated basin, using discharge measurements at main canal offtakes and satellite imagery. The accuracy of the computed canal water distribution was highest when Landsat image (with a resolution of 30 m) was used to identify the shape of the irrigated areas. This implies that high-resolution satellite images can be used to discern canal water use from groundwater use.Net groundwater use was computed for the entire Rechna Doab using the derived maps of spatially distributed water balance components. On an annual basis, an average net groundwater use of 82 mm yr <sup>-1</SUP>was estimated, which coincides with groundwater inflows of 53 mm yr <sup>-1</SUP>that conserve the water balance of the saturated zone. Using the technique presented in this thesis, the computed net groundwater use, based on an unsaturated zone water balance, deviates from classical method estimates, using specific yield and fluctuation of the phreatic surface, by 65%. The deviation from estimates using tubewell withdrawal related data is even higher (several hundred percent).