"Multi-Temporal UAV Multispectral Dataset for Solar Illumination-Aware Vegetation Index Analysis in Paddy Rice Fields"

"This dataset contains multi-temporal UAV (Unmanned Aerial Vehicle) multispectral imagery-derived measurements collected over paddy rice (Oryza sativa) fields in West Bengal, India, during the 2025 Kharif cropping season (August\u2013October 2025). Data were acquired using a DJI Mavic 3M multispectral UAV across two geographic locations , Debra and IIT Kharagpur, spanning three distinct field plots (Field 1, Field 2, Field 3) at flight altitudes ranging from approximately 2.7 m to 30 m above ground level.The dataset comprises 3,901 frame-level observations across 7 flight dates, capturing the full paddy growth trajectory from vegetative to reproductive stages. Each record includes raw digital number (DN) reflectance values, empirically calibrated surface reflectance, vegetation indices (NDVI, NDRE, GNDVI) computed at both DN and reflectance levels, UAV flight attitude parameters (yaw, pitch, roll: both flight body and gimbal), radiometric calibration metadata (radiometric scale, black level), and solar geometry parameters (solar zenith angle, solar azimuth angle, and cosine of zenith).A key feature of this dataset is the explicit inclusion of per-frame solar geometry (solar zenith and azimuth angles computed from GPS timestamp and coordinates), enabling rigorous investigation of solar illumination-induced radiometric bias in UAV-borne vegetation indices. The dataset supports research into illumination normalisation, cosZ\u2013NDVI confound analysis, cross-date radiometric harmonisation, and the disentanglement of spatial versus illumination effects in lawnmower flight patterns. Flights with altitudes \u2264 4 m have been excluded to ensure stable aerodynamic conditions."