A simple model for yield prediciton of rice based on vegetation index derived from satellite and AMeDAS data during ripening period

Abstract The present study was conducted to show a simple model for rice yield predicting by using a vegetation index (NDVI) derived from satellite and meteorological data. In a field experiment, the relationship between the vegetation index and radiation absorbed by the rice canopy was investigated from transplanting to maturity. Their correlation held. This result revealed that the vegetation index could be used as a measure of absorptance of solar radiation by rice canopy. NDVI multiplied by solar radiation (SR ) every day was accumulated (sum(SR · NDVI)) from the field experiment. sum(SR · NDVI) was plotted against above ground dry matter. It was obvious that they had a strong relationship. Rice yield largely depends on solar radiation and air temperature during the ripening period. Air temperature affects dry matter production. Relationships between YSR sup (-1) ( Y : rice yield, SR : solar radiation) and mean air temperature were investigated from meteorological data and statistical data on rice yield. There was an optimum air temperature, 21.3 deg C, for ripening. When it was near 21.3 deg C in the ripening period, the rice yield was higher. We proposed a simple model for yield prediction of rice based on these results. The model is composed with SR · NDVI and the optimum air temperature. Vegetation index was derived from 3 years. LANDSAT TM data in Toyama, Ishikawa, Fukui and Nagano prefectures at heading. The meteorological data was used from AMeDAS data. The model was described as follows : Y=0.728SR · NDVI 2.04(T-21.3) sup 2+282 (r sup 2 0.65,n 43) where Y is rice yield (kg 10a sup (-1)), SR is solar radiation (MJ m sup (-2)) during the ripening period (from 10 days before heading to 30 days after heading), T is mean air temperature (deg C) during the ripening period. RMSE was 33.7kg 10a sup (-1). The model revealed good precision.