Correcting Thermal Drift: A Field-Validated Model for UAV Surveys in Agriculture

Thermal infrared (TIR) imaging from UAVs is increasingly employed in precision agriculture to monitor spatial and temporal variations in crop surface temperatures. However, absolute temperature values retrieved from UAV-mounted thermal sensors are affected by environmental and geometric factors, which may limit their reliability for quantitative analyses. This contribution presents a model-based temperature correction framework that integrates radiometrically converted thermal data with ancillary information (i.e., ambient temperature, relative humidity and surface emissivity) collected during UAV flights. The method estimates corrected absolute temperatures and quantifies associated uncertainty, enhancing the robustness of temperature maps across varying operational conditions. The approach is applied to a dataset acquired in a vineyard in Montemurlo (Italy), by performing multiple UAV flights under different atmospheric conditions. Results confirm the potential of the proposed correction model for improving temperature accuracy and enabling uncertainty-aware thermal mapping.