Geo-statistical models for determining spatial variation and spatial dependency of soil arsenic in Bangladesh

Arsenic (As) toxicity has become a major problem since a few decades in different parts of Bangladesh mainly due to the massive groundwater extraction for both drinking and irrigation purposes. This study was undertaken to investigate the spatial variation and spatial dependency of soil As at four locations in Bangladesh. Composite soil samples were collected at 0 ‒ 45 cm depth in a grid area of 1 m2 to conduct the study of a shallow tube well command area (micro level). Based on composite soil sampling in four different study locations, laboratory analysis and statistical modelling, the spatial variation and dependency of soil As was estimated. Four study area maps were digitised using ArcGIS. Arsenic concentration varied widely within the study areas. The extent and propensity of As concentration was higher in areas with a high concentration of As in groundwater and soil. About 15.01 ppm of As is loaded into soils through groundwater irrigation every year in the As affected areas. The semi-variogram model for describing spatial dependency of soil As was found to be scale dependent. At study area scale, the appropriate model was identified as spherical. Kriging method appeared to be more suitable to create an interpolated surface for study area scale.  Development of As tolerant rice varieties, promoting cropping  patterns that require less irrigation water, and alternate wetting  and drying (AWD) method can be remedial measures to reduce  As contamination in soil as well as in the food chain.