Historical wheat yield mapping using time-series satellite data and district-wise yield statistics over Uttar Pradesh state, India

Abstract

Crop productivity has often been affected by undesirable climatic events such as heat stress, flood, unseasonal rainfall, drought, etc. Historical crop yield estimation and mapping shall provide a pivotal input for food security measures and planning purposes. The present study deployed various satellite (i.e., normalized difference vegetation index) and gridded (i.e., precipitation, temperature, evapotranspiration) products to portray the historical wheat yield at satellite pixel-level during 2001–2019 over the Uttar Pradesh state of India. Directorate of Economics and Statistics (DES) based district-wise wheat yield dataset was also used with satellite variables to develop the historical wheat yield model based on the multiple regression analysis. Spatially, higher, lower, and lower to moderate wheat yields are observed in the western and northwest parts, southern and southeast parts, and lower-middle to the lower right parts of the state respectively throughout the study years. Remarkably, the years 2017,2018, and 2019 witnessed higher yields, while 2004, 2006, 2007, 2008, and 2015 witnessed lower yield. The developed wheat yield models performed well, as R2 values were observed between 0.3 and 0.76 between the satellite-derived and DES-based district-wise mean yield. The mean absolute error was found to be considerable and ranged between 0.22 and 1.7 t/ha during the study years. The adopted methodology can be used across the different parts of the globe at the local to country scale for a quantitative crop yield depiction and mapping at the satellite pixel level. Such historical crop yield mapping can assist in understanding the long-term crop yield pattern in response to the climate change activity, and hence food security measures can be formulated and implemented.