Quantifying Regional Groundwater Storage Changes in High Mountain Asia by Integrating In-Situ Groundwater Levels with Remotely Sensed Data from GRACE Satellites

Abstract:

Regional groundwater storage changes can be estimated either through large-scale remote sensing approaches or by upscaling groundwater level measurements observed at point scale. The former approach requires the disaggregation of the full water budget to isolate the groundwater storage component and is limited by the reliability of the independent models used and/or observations of the other components. The latter approach requires storage coefficients that convert discrete water levels to storage change which are then upscaled to the region of interest. The upscaling process introduces errors and, in particular, is limited by the storage coefficients used as they are traditionally estimated at point scale and likely not representative at a larger scale. Therefore, we propose a method to calculate regional groundwater storage changes. For this, we constrain large scale aquifer storage parameters by the combined use of in-situ groundwater level data and remote sensing observations (GRACE) of changes in total water storage (TWSA) in the High Mountain Asia region covering eastern Pakistan, north/ northwestern India, and Bangladesh.

We use a unique secondary dataset of measured groundwater levels in more than 12,000 wells with adequate spatial and temporal coverage across the study area. After processing the dataset to ensure quality control, we filter the observations by layer type (shallow vs. deep) and use a kriging interpolation scheme to upscale changes in groundwater levels (GWLA) to the GRACE scale, i.e., 1^o x 1^o grid. Since GRACE TWSA trends from Jan 2003- July 2016 are dominated by trends in groundwater storage changes (GWSA) for this region, we estimate large-scale aquifer storage parameters by minimizing the residuals between in-situ GWSA trends and GRACE trends using a gradient-based inversion technique. We find the total GWSA trend for the study area falls well within the reported GRACE trend of −23.2 ± 4.3 Gt yr⁻¹. The optimized storage coefficients show an excellent correlation (0.98) between aggregated in-situ GWSA trend and GRACE trend for the regional aquifers.