Sustainable crop intensification through surface water irrigation in Bangladesh? A geospatial assessment of landscape-scale production potential

[Display omitted] •Arable land area in South Asia is declining while cereal demand is increasing.•Bangladesh emphasizes surface water irrigation (SWI) for sustainable intensification.•Remotely sensed, geospatial, and farmers’ yield data were integrated to target SWI.•20,800 and 103,000ha of fallow and rainfed cropland could benefit from SWI.•SWI policy could substantially increase maize and wheat, but not rice production. Changing dietary preferences and population growth in South Asia have resulted in increasing demand for wheat and maize, along side high and sustained demand for rice. In the highly productive northwestern Indo-Gangetic Plains of South Asia, farmers utilize groundwater irrigation to assure that at least two of these crops are sequenced on the same field within the same year. Such double cropping has had a significant and positive influence on regional agricultural productivity. But in the risk-prone and food insecure lower Eastern Indo-Gangetic Plains (EIGP), cropping is less intensive. During the dryer winter months, arable land is frequently fallowed or devoted to lower yielding rainfed legumes. Seeing opportunity to boost cereals production, particularly for rice, donors and land use policy makers have consequently reprioritized agricultural development investments in this impoverished region. Tapping groundwater for irrigation and intensified double cropping, however, is unlikely to be economically viable or environmentally sound in the EIGP. Constraints include saline shallow water tables and the prohibitively high installation and energetic extraction costs from deeper freshwater aquifers. The network of largely underutilized rivers and natural canals in the EIGP could conversely be tapped to provide less energetically and economically costly surface water irrigation (SWI). This approach is now championed by the Government of Bangladesh, which has requested USD 500 million from donors to implement land and water use policies to facilitate SWI and double cropping. Precise geospatial assessment of where freshwater flows are most prominent, or where viable fallow or low production intensity cropland is most common, however remains lacking. In response, we used remotely sensed data to identify agricultural land, detect the temporal availability of freshwater in rivers and canals, and assess crop production intensity over a three-year study period in a 33,750km2 case study area in southwestern Bangladesh. We combined these data with geor