Groundwater use and rapid irrigation expansion in a changing climate: Hydrological drivers in one of the world’s food bowls

•Water balance (1981–2015) in northwest Bangladesh to assess groundwater level falls.•Two evapotranspiration methods yield similar results capturing irrigation increase.•Trends show falls in rainfall and groundwater, evapotranspiration remains similar.•Soil permeability and groundwater recharge limit irrigation supply from soil stores.•Many factors besides irrigation expansion contribute to groundwater level falls. In recent decades, increased groundwater use enabled a large areal increase in irrigated dry season crops in northwest Bangladesh. Concurrent declining groundwater levels across the region are of great concern for food security. A water balance model approach that considered changes in irrigated agriculture was implemented to assess changes over the long-term to three five-year evaluation periods (1985–1989, 1998–2002 and 2011–2015) and seasonally (annual, dry season and wet season). The model used two different methods that explicitly capture changes in irrigation to compute and compare actual evapotranspiration (ETa). The first method used MODIS satellite data to estimate a crop coefficient based on vegetation indices (at 500 m spatial resolution) scaled by reference crop evapotranspiration (ETref). The second method used a crop coefficient approach based on survey data of crop areas at the district level and subsequently scaled by ETref. Both methods yielded very similar results at the district level, with correlation coefficients between 0.75 and 0.89. The maximum difference between the monthly averages was only of 5.4%. Notwithstanding the observed overall increase in irrigated areas, estimated overall mean annual ETa is similar through the analysis (ca. 1100 mm) and district-level trends were mixed (some increasing and some decreasing) showing in some districts a weak association between ETa and the related declining groundwater level. From the water balance, it is inferred that both the groundwater extraction (by pumping for irrigation and capillary rise to supply roots) and the groundwater recharge reduced from 1998–2002 to 2011–2015, and with deeper groundwater in later years, much of the irrigation water supply is by soil water storage rather than by groundwater. As highlighted in this paper, there are other factors aside crop expansion that may have contributed to the groundwater decline, thus a single policy or management change such as restricting groundwater extraction for irrigation may alone be inadequate to reverse declining gro