Reliable NDVI estimation in wheat using low-Cost UAV-derived RGB vegetation indices

The implementation of precision agriculture techniques requires the presence of decision support systems. However, the technological penetration in small-medium size farms is usually low. Though, remote sensing techniques for monitoring crops status, founds limited diffusion for the cost of those practices. Therefore, the identification of low-costs indices based on RGB images, to obtain comparable information of multispectral parameters such as NDVI, can broaden the range of farms in order to integrate precision agriculture paradigm. In this study, several indices based on RGB images obtained by commercial UAV, were correlated with NDVI values in common wheat, from tillering to grain filling phases. Wheat cultivation took place at mesocosm level, in growth boxes placed in open air. Moreover, the degree of correlation between RGB indices and NDVI was tested at three ground resolutions (1.5 mm pixel-1, 20.0 mm pixel-1, 30.0 mm pixel-1). Of twelve indices tested, eleven were correlated with NDVI, and five of those, (DCGI, GRVI, SAVIgreen, VARI and GMR) showed a high degree of correlation (Adjusted R2>0.70). Additionally, the decrease of degree of correlation between RGB indices and NDVI was scarcely influenced by ground resolution. Though, the use of indices with low computation requirement, obtained by raw images derived from commercial low-cost instruments, can substantially widen the access to precision farming approach.