Practical Method for Estimating Rice Growth Using Vegetation Index Based on UAV Remote Sensing

Abstract

In this study，five normalized vegetation indices （NDVI，BNDVI，GNDVI，NDRE，and LCI） were calculated for three rice varieties （Koshihikari, Kinuhikari, and Hinohikari） from the reflectance of multispectral images captured using an unmanned aerial vehicle（UAV）.The relationship between growth （the product of plant length，number of tillers per m²，and SPAD value in rice） and corrected rice growth （a parameter calculated by multiplying growth with the effective integrated temperature） was analyzed for different years （2023 and 2024），regions，planting densities，nitrogen fertilization conditions，and field characteristics.Consequently，the coefficient of determination （R²） of the regression equation was higher for all normalized vegetation indices with corrected growth than for those with growth in the combined 2023–2024 dataset. Furthermore，NDRE and LCI，two normalized vegetation indices that use reflectance in the red-edge wavelength range，had similarly high R² values in the regression equations for all species.The regression line between the vegetation index and growth for some cultivars deviated significantly from year to year；however, multiplying growth with the effective integrated temperature reduced the differences between years.Thus，in relation to the vegetation index， the effective integrated temperature corrects for differences in weather conditions that affect growth.An estimation can be made using a constant equation，regardless of the year.This is believed to have improved practicality in the field.Conversely，the regression equation between the vegetation index and corrected growth differed among the species，indicating that the regression equation should be applied to each species.