A UAV-Proxied Satellite Remote Sensing Approach for Winter Wheat Plant Nitrogen Concentration Mapping

Accurate and real-time prediction of plant nitrogen concentration (PNC) is crucial for assessing the N status of winter wheat. Satellite remote sensing enables large-scale PNC monitoring but suffers from coarse spatial resolution, causing scale mismatches with field sampling. Unmanned aerial vehicles (UAV) remote sensing bridges this gap, linking satellite observations with ground measurements. This study mapped regional-scale winter wheat PNC using ground-measured PNC, UAV (DJI Phantom 4 Multispectral), and satellite (PlanetScope, Sentinel-2) data. First, UAV-based vegetation indices (VIs) with machine learning predicted PNC at the field scale. Next, UAV data were aggregated to match satellite resolution, producing field-scale PNC maps. These were then used as regional-scale samples, combined with satellite VIs, to develop PNC estimation models. At the regional scale, support vector regression with Sentinel-2 VIs performed best (R² = 0.95, RMSE = 0.09%, RPD = 4.70). Validation with ground samples and UAV-scale PNC maps confirmed the reliability of cross-scale models (R² = 0.75–0.96, RMSE = 0.09–0.29%, RPD = 1.93–4.85). Among the two satellite sensors evaluated, Sentinel-2 outperformed PlanetScope in PNC estimation, likely due to its more optimal red-edge spectral configuration. These findings highlight the effectiveness of UAV-satellite cross-scale remote sensing for regional-scale PNC mapping, providing a valuable tool for large-scale crop N monitoring.