Abstract:

Leaf area index (LAI) is an important parameter in crop growth status monitoring and yield forecasting. A vegetation index (VI) based on spectral reflectance measurements has been proposed as a reliable nondestructive method for quickly estimating LAI. To determine the best broadband index for estimating LAI in rice (Oryza sativa), field canopy reflectance values were measured over the whole rice growth cycle using a portable multi-spectral radiometer, and LAI were simultaneously determined by destructive sampling. Several vegetation indices such as normalized difference vegetation index (NDVI), ratio vegetation index (RVI), soil-adjusted vegetation index (SAVI) and the like were derived from these spectral measurements and their correlation with respect to LAI quantified. Also, their relative predictive powers were estimated by comparing determine coefficient (R2), root mean square error (RMSE) and precision and accuracy. The results showed that the power of VI for LAI assessment was the best during vegetative growth, and mainly depended on the range of variation in the experimental data. Vegetation indices accurately tracked changes in LAI when data were analyzed across a broad range of different growth stages and nitrogen levels. RVI, RDVI, and R810/R560 showed a power relation with LAI, while NDVI, PVI, DVI, SAVI and TSAVI showed an exponential relation. R810/R560 produced the best estimate of LAI among these indices. The regression equation was tested by independent datasets and the estimation accuracy was about 91.22% with RMSE of 0.480 5 and average relative error of -0.013. The results indicated that LAI monitoring in rice by means of the ratio index of near infrared band to green band from broadband spectral signatures appears very promising.