Abstract
The growing demand for rice products highlights the importance of maximizing yield and closely monitoring crop development. In this context, spectral vegetation indices (VIs) play a key role in characterizing plant growth. This study evaluated the performance of 20 vegetation indices derived from multi-temporal Sentinel-2 imagery for estimating the Leaf Area Index (LAI) in rice crops. In-situ LAI samples were collected at various phenological stages from selected regions in Bulgaria and Türkiye, in coordination with field campaigns conducted in parallel with Sentinel-2 image acquisition. Among the indices, GOSAVI showed the highest Pearson correlation with LAI (r = 0.74). The Random Forest algorithm was employed to estimate LAI from each index, with SAVI yielding the highest accuracy (R2 = 0.59, RMSE = 1.4). The findings indicate that it can effectively support practical and efficient LAI estimation with Sentinel-2 data for data collected from different regions and help improve rice crop monitoring studies.