Numerical Groundwater Flow Modeling of the Arsenic Contaminated Gotra Aquifer, West Bengal, India

We present results of a groundwater flow study in a 7 square kilometer region centered on our field site in the village of Gotra, West Bengal, India, where naturally occurring arsenic contaminates shallow groundwater (mostly < 40 m below surface) at levels that exceed World Health Organization limits. Field investigations since 2004 include sediment coring, geochemical surveys of domestic wells and dedicated piezometers, and monitoring of piezometric levels with dataloggers in several wells over a period of three years. The village itself is situated upon the natural levee of a sinuous abandoned channel, which terminates a classical fluvio-deltaic depositional sequence originating from the north east. The formerly prograding meander bend deposited point bar sands that now comprise the 25-30m-thick shallow aquifer, while incising a deeper sand unit as well as a floodplain sequence to the south west. Hand-pumped domestic tubewells are restricted to the point bar sands, whereas the majority of irrigation wells are screened within the deeper aquifer. Both steady-state and transient simulations were used to estimate aquifer recharge and boundary fluxes, as well as to distinguish groundwater flow paths and average residence times under the influence of irrigation and seasonal stresses. The hydrogeology of the deeper aquifer, as well as the thickness and continuity of the low permeable paleosol separating shallow and deep flow systems are major uncertainties that are explored through model sensitivity testing. Recharge to the shallow flow system is estimated at ~170mm/yr, predominantly by monsoon precipitation. Groundwater in the arsenic affected area generally flows downward through the fine-grained channel-fill deposits, then horizontally through the point bar sands perpendicular to the trend of the abandoned channel. This flowpath coincides with observed geochemical gradients from surveys of domestic well quality conducted in previous years, and supports a model of arsenic release from the low permeability unit. Average residence times in the point bar sands and channel-fill deposits are estimated at 50 and 310 years, respectively, and are highly sensitive to variations in recharge, shallow aquifer conductivity, and hydraulic conductivity of seepage layers that line surface water bottoms.