Simulation of water productivity of wheat in northwestern Bangladesh using multi-satellite data

Water productivity (WP), a performance indicator, defines if the agricultural production system uses the resources competently. Spatial information on crop WP is fundamental to knowing where a gap in WP exists along with underlying reasons within the gaps, where a crop production system performs well, and where developments are still possible to enhance the water-productive performance. This study proposes to assess spatial wheat WP in drought-prone northwestern Bangladesh. Both primary and secondary data, such as Sentinel 2 MSI, Landsat 8 OLI satellite images, actual evapotranspiration (ETa) from Lysimeter, and wheat farmers’ field data over the study area, were collected to accomplish this study. The wheat-cultivated fields were delineated first using Sentinel 2 MSI data during the 2018 wheat growing season. WP parameters, namely crop yield and ETa were estimated using Landsat 8 satellite data. Based on yield data from the farmer's field and normalized difference vegetation index (NDVI) values generated from the same fields, a linear regression model was developed. This model was applied to obtain crop yield maps from the NDVI maps of the study area. A remote sensing-based well-tested, robust algorithm surface energy balance algorithm for land (SEBAL) was used to estimate ETa using Landsat 8 images and local ground information. Cloud-free single-date images corresponding to wheat growth stages such as initial, development, mid-season, and late-season were used to estimate seasonal ETa. SEBAL-estimated single date ETa was validated by newly developed (under this study) Lysimeter-measured ETa. Wheat maps were then masked to get crop-specific yield and ETa maps. Finally, the WP maps of wheat were produced by the mathematical operation of the yield and ETa raster maps. The NDVI- crop yield (Y) models were calibrated and validated, which shows an acceptable degree of dispersion between observed and estimated values. Thakurgaon and Panchagorh districts (second administrative tier) are higher wheat-yielding districts in the study area. The average wheat-yield of the study area was 3.51 t ha−1, which is close to the department of agricultural extension (officially) reported data. In the case of the denominator (ETa) of WP estimation, results show that SEBAL-estimated wheat ETa on the image dates is close to ETa estimated by the Lysimeter. The average seasonal ETa for wheat is found as 253 mm. The estimated highest and lowest ETa demands are found in Thakurgaon and Panchagorh districts, respectively. The estimated WP of wheat ranged between 1.33 and 1.46 kg m−3, with an average of 1.39 kg m−3 among the eight study districts. Panchagorh and Thakurgaon districts are the highest water productive, whereas Gaibandha and Nilphamari show the minimum level of the WP. Among the Y-ETa, WP-ETa, and WP-Y relationships, WP-Y shows the highest correlation (r2 =0.73), as the higher WP observed mainly due to high yielding capacity. Besides, suitable soil characteristics and weather conditions favor wheat WP in those areas. This study identifies high-performing bright and low-performing hot spots of WP across the northwest of Bangladesh. Finally, stakeholders and decision-makers can use the information to define priority areas, set goals for improvement, and justify the type of investment or measures for improving water-productive agriculture in Bangladesh.