Evaluation of soil-dependent crop yield outcomes in Nepal using ground and satellite-based approaches

Abstract

Abstract Smallholder farmers face many constraints to achieving food security. Optimal soil management is often limited by a lack of accessible and accurate soil characterization, and an associated lack of soil-specific management practice recommendations. Crop yields depend on both soil quality and soil-mediated fertilizer responses. Existing research on soil-fertilizer interactions is primarily based on farm trials and/or survey data, which are resource intensive and typically restricted to local scales. High-resolution (∼10-meter) remote sensing data and digital soil maps provide a low cost, scalable alternative. Here, we deploy methods based on the Sentinel satellite constellation to estimate soil and fertilizer impacts on irrigated wheat grain yields in Nepal and to inform precision soil and nutrient management recommendations. We first combine field data with Sentinel-1 and Sentinel-2 imagery to delineate wheat cropping areas for 2016–2019 with 92 % accuracy. We then estimate wheat yields at 10-meter resolution using Sentinel-2 and weather covariates based on yield models parameterized from two different methods: 1) APSIM crop model simulations and 2) ground cropcuts from 147 fields. Using a large dataset of soil samples collected by the Nepal Agricultural Research Council, we examine the linear and non-linear effects of soil properties on wheat yields. Finally, the soil maps were combined with a survey of field-level crop management data and our yield estimates to test the interaction of soil quality with fertilizer effectiveness. Our ground-calibrated satellite model predicted yields with good accuracy (R2 = 0.55), while the uncalibrated simulation-based approach had weaker but significant prediction accuracy (R2 = 0.24). We find statistically significant gains in yield of 0.9–2.4 % are possible by increasing soil organic matter and zinc from highly deficient values to optimal values of 2.2 % organic matter (OM) and 0.67 ppm zinc (Zn). Using digital soil maps of Nepal produced by the International Maize and Wheat Improvement Center (CIMMYT), we map croplands deficient in zinc (Zn