Evaluating Mechanically-caused Crop Damage Using Two Simple UAV-based Assessment Techniques

Abstract. The increasing frequency of hydrometeorological extremes, such as torrential rainfall, strong winds, and hailstorms, often causes widespread mechanical damage to crops. This study evaluates the potential of cost-effective unmanned aerial vehicle (UAV) photogrammetry with a standard RGB camera for quantifying crop damage. A maize field with mechanical damage caused by wild boar activity was used as an analogue for storm-induced damage. Two approaches were applied: (i) a 3D structural method based on Canopy Surface Models (CSMs) derived from Structure-from-Motion (SfM) photogrammetry, and (ii) automated image classification using a Support Vector Machine (SVM) combined with Object-Based Image Analysis (OBIA). The accuracy of the damage assessment was compared using two terrain inputs: a UAV-derived DEM (UAV DEM) and the official Czech national LiDAR-based DEM (DEM 5G). The results showed high consistency between both methods and datasets. The relative crop damage rate was 29.25% with the UAV DEM and 26.76% with the DEM 5G, with a spatial agreement exceeding 95%. Jaccard similarity coefficients confirmed strong concordance (0.8953 and 0.9207). The findings highlight the applicability of UAV-based 3D structural analysis for late-stage crop monitoring, when spectral indices lose reliability. They also emphasise that the official DEM 5G can serve as a suitable substitute for a UAV-derived DEM in damage assessment. The methodology thus represents a rapid, cost-effective, and operationally feasible solution for agricultural monitoring, insurance claims, and environmental management.