TRANSIENT ARSENIC CONCENTRATIONS IN A SHALLOW AQUIFER OF BANGLADESH INDUCED BY IRRIGATION PUMPING

Little is known about the stability of arsenic (As) concentration in shallow (<30 m) aquifers of the Bengal basin on decadal time scales even though many households have switched to neighboring low-As well as a way of reducing their exposure. To better understand the temporal variability of groundwater As, we have compared three well testing campaigns within a 25 km² area of Araihazar, Bangladesh accomplished over the last 14 years. Over 4,500 shallow wells were sampled during the first blanket survey in 2000-01 and were analyzed by GFAA-spectrometry. In the second blanket survey of 2012-13, 8,200 wells were tested with a field-kit including those still in existence from 2000-01. Due to the high rate of replacement of wells, only 350 shallow wells sampled in 2000-01 could be identified with certainty during the third survey in 2014 and were analyzed by ICP-MS. Comparison of a total of 308 samples collected in 2014 with the archived 2000-01 samples reanalyzed by ICP-MS shows an overall decline of mean As concentration (by 10%) in the area, but more interestingly As concentrations at the higher initial levels have decreased whereas As concentrations at initially lower levels have increased over time. The majority of the samples (64%) indicate statistically significant (2σ) change in As concentrations with increases ranging from 8 to 350 ppb and decreases in the 8-310 ppb range. In order to make a comparison based on the two blanket surveys, each kit reading of 2012-13 was calibrated using 940 quality control samples analyzed by ICP-MS. The test results from thousands of mostly different wells in the area were then averaged over 83 blocks (600X600 m² areas). Statistical analysis of the results indicates a significant (1σ) decrease in average As concentrations in 27 out of 83 blocks and a significant increase in 7 blocks. The trends are consistent with the observations based on the much smaller number of resampled wells and indicate a gradual homogenization of As concentrations that accompanied a decrease in the areal average from 115 to 100 ppb. We use a simple box-model to show that the observed trend can be explained by mixing of As across blocks at the rate set by irrigation pumping in the study area. The long-term implication is that households currently relying on shallow low-As wells to lower their exposure may not be able to do so indefinitely.