Use of Vegetation Indices Derived from Unmanned Aerial Vehicles (UAVs) Imagery for Monitoring Growth and Yield of Paddy and Identifying Weeds in Paddy Fields: A Case Study in Sri Lanka

The use of drone-based imagery enables data-driven decision making, by optimizing farming practices to enhance crop yields. This study was conducted to evaluate various vegetation indices for monitoring the growth of rice plants and detecting weeds in paddy fields. Field experiments were conducted at the Rice Research and Development Institute (RRDI) at Batalagoda [Agroecological region (AER): IL1] in Sri Lanka during the 2021 Yala cultivating season (March to September). Five rice varieties, namely, Bg 250, Bg 300, Bg 359, At 362, and Bg 450, along with three weed species (Echinochloa crus-galli, Ischaemum rugosum, Cyperus iria) were included in the study. Data was collected using a DJI Phantom 4 Multispectral drone, with flights conducted at seven-day intervals. Image analysis was performed using PIX4Dfields software. A positive correlation was observed between the Normalized Difference Vegetation Index (NDVI) values and the paddy yield, highlighting the potential of NDVI as a reliable indicator for monitoring rice plant growth. Further, the modified Excess Green (ExG) Index in Red-Green mode successfully identified the presence of weeds at 14 days after sowing (DAS). The Weed Detection Index (WADI) in both spectral and Red-Green modes proved most effective for identifying weeds at 55 DAS. The findings underscored the utility of precision agriculture (PA) technologies in paddy farming, particularly for weed detection. The application of these indices can contribute significantly to the development of site specific weed management strategies, enhancing the sustainability and efficiency of paddy production in Sri Lanka.