Abstract：The development of UAV (unmanned aerial vehicle) hyperspectral remote sensing technology provides a new means for accurate and high-throughput observation of field crops. In rapeseed planting management， the accurate determination of moisture content in rapeseed silique peel is of great significance to determine its silique shattering resistance and optimal harvest period and reduce yield loss. However， the traditional field survey sampling work is lagging and subjective， and it is difficult for large-scale implementation and of low efficiency. Therefore， this study used the UAV hyperspectral sensor to obtain canopy reflectance data of rapeseed at pod stage in the study area， and constructed prediction models of moisture content in rapeseed silique peel. The results showed that the hyperspectral vegetation indices related to chlorophyll and vegetation senescence performed well in predicting moisture content. Based on this， the random forest regression model was constructed with R2of 0.75 and RMSE of 1.67% on the test set. This study highlights the advantages of UAV hyperspectral data in predicting the moisture content in rapeseed silique peel， and provides a powerful reference for accurately predicting moisture content and assisting rapeseed harvest decisions.
2025, Vol. 59 
