Spatial and temporal modeling of regional groundwater level in context of climate change

Abstract

The accurate prediction of groundwater resources as the sole source of drinking and irrigation based agriculture in the northwestern part of Bangladesh is important for the sustainable use and management of this already stressed precious resource.Groundwater level data collected from 130 sites across 25 Upazilas (sub-district) of three northwest districts of Bangladesh were used in this study to access the impacts of climate change on groundwater resources in the region. Several geostatistical and determistic interpolation methods as well as data mining techniques such as, Support Vector Machines (SVM) and Artificial Neural Network (ANN) were investigated for spatial and temporal modeling of groundwater level. The study revealed that co-kriging gives the best estimation of spatial distribution of water table when soil infiltration information is provided. On the other hand, Artificial Neural Network (ANN) was found to model groundwater table fluctuation more accurately compared to other data mining approaches. Therefore, ANN was used to project the changes in groundwater level under projected climate data obtained through statistical downscaling of global circulation model outputs. Groundwater drought situations during base year and under projected climate were investigated using the Cumulative Deficit approach in a geographical information system. The study revealed that groundwater scarcity in at least 27% of the study area will be an every year phenomenon in the region due to climate change. Analysis of climate change and groundwater hydrographs reveals that no appreciable change in precipitation, but increases in temperature as well as increase in groundwater extraction for irrigation in the dry season are the causes of groundwater scarcity in the region.