Data fusion approach for map-based variable-rate nitrogen fertilization in barley and wheat

Abstract

Nitrogen (N) is a main parameter affecting crop growth and yield. However, when variable rate N fertilization (VRNF) is adopted, a main question arises of whether to add more on the highly fertile zone or on the less fertile zone. This study aimed to compare two map-based VRNF treatments, e.g., variable-rate 1 (VRNF1) with the application of recommended rate (RR) of N fertilizer plus 40-50% to the high fertility zones, RR to the medium-fertility zones and RR minus 40-50% to the least fertile zones, and VRNF2 following the opposite scenario to that of VRNF1. Recommendations for both VRNF treatments were derived from the fusion of soil data and crop data, e.g., normalized difference vegetation index (NDVI) and historical yield and were compared with traditional uniform rate (UR) application. Online soil measurement was carried out in two fields in Belgium, using a multi-sensor platform that collects high-resolution data on key soil properties using a fiber type visible and near-infrared (vis-NIR) spectrophotometer. The vis-NIR sensor was calibrated using a partial least squares regression (PLSR) to predict pH, organic carbon (OC), moisture content (MC), calcium (Ca), sodium (Na), magnesium (Mg), phosphorous (P) and potassium (K). The predicted soil properties were fused with NDVI and historical yield data to delineate management zone (MZ) maps with different fertility classes. A cost-benefit analysis was carried out using the cost of fertilizer applied as input cost and the price of barley and wheat in 2019. Results showed that the VRNF2 treatment was more profitable compared to both VRNF1 (11.55 and 19.52 EUR ha-1) and UR (16.40 and 148.78 EUR ha-1). The VRNF2 treatment maintained or increased crop yield by up to 10.4% and reduced the amount of N fertilizer applied by up to 19.4%, compared to the traditional UR. Because of the lower amount of N fertilizer used compared to the other two treatments, the VRNF2 would reduce the negative environmental impact by reducing the amount of N fertilizer applied. Therefore, we recommend the VRNF2 based-on data fusion of soil and crop parameters as the best approach to manage N fertilization site specifically.