Estimating rice crop (Oryza sativa L.) parameters during the 'Yala' season in Sri Lanka using UAV multispectral indices

Abstract

Remote estimation of aboveground biomass (AGB), plant height (PH), and rice yield (RY) before the actual harvest is vital for the early management of rice crops. Traditionally, on-ground spot measurements and destructive sampling methods were conducted across the field which is problematic due to poor temporal and spatial representation, time consumption, delayed crop management and inefficient input application. Thus, this study evaluates the advanced aerial multispectral imagery technique to explore the association of image-derived vegetation indices (VIs) with the ground measured AGB, PH and RY at crop season “Yala” 2021 in the controlled field and extend the relationship with the commercial farmer field data. The experiment was carried out at a controlled rice field and then at a commercial farmer field cultivated with three-month crop duration rice varieties as Bg 300 and Bg 374, respectively. Multispectral images of the field were acquired by an Unmanned Ariel Vehicle (UAV, DJI P4-multispectral, Shenzhen, China) at 25m flying height during the booting stage of the rice plant. Then VIs was derived using QGIS (version 3.16) software and analyzed to select the best combination of VIs explaining the ground measured parameters using partial least squares regression (PLS-R). According to the results, a strong agreement between growth parameters during the booting stage at 25m flight height was proven. When comparing the relationship at the commercial farmer field, the estimated AGB showed strong accuracy (R2 = 0.62, RMSE = 0.055 kg/m2) with the observed AGB using VIs (GRVI and SR). Estimated PH showed weak accuracy (R2 = 0.04, RMSE = 19.33 cm) with the observed PH using (CI red edge, TCARI and NPCI) and estimated RY showed strong accuracy (R2 = 0.75, RMSE = 817.8 kg/ha) with observed RY using VIs (TVI, SR and MSR). Therefore, this study proves an association of VIs with rice crop parameters AGB and RY in rice cultivation that can be used to measure non-destructively prior to harvest during “Yala” in Sri Lanka.