Combining Crop Growth Model And Remote Sensing Data For Rice Growth Monitoring And Yield Prediction

Accurate and up-to-date information on rice growth and yield prediction always is a vital subject that needs governments and scientific research organization to solve. Yield prediction using RS data is objective,quantitative and macroscopic, but it usually adopts statistic model that leads to instability of precision on forecasting. Crop growth models describe crop growth as a function of time and have definite meaning of biology and physics. But crop growth models only select representative spots to simulate. Monitoring large scale crop growth and simulating yield that is to reflect the feature of distribution in area is difficult. This article combined RS technique with rice growth model (RCSODS) to concurrent processing the problem about time course of "point" and space distribution of "surface".This article combined RS data (NOAA/AVHRR, LANDSATTM) with rice growth model to establish remote sensing simulation model. On the basics of the relation NDVI and LAI, RCSODS was modified to accept observed LAI at certain times during the season and simulated the yield. The main results of research are as follows:1 ,Using software we corrected NOAA images according to TM image and got accurately integration result.2, Using the visible channel and near infrared channel of NOAA satellite we built vegetation index NDVI, based on the relation between NDVI and LAI, large scale LAI of rice which effectively represents rice growth status can be conversed. Through this method we monitor rice growth.3 , We compared actual yields with simulated yields of four pixels of sample using different temporal NOAA data. the average accuracy was over 90%; by comparing the simulated yields, we found: when we simulated the yields using single temporal RS data selecting the time from enclosed line to florescence can get higher accuracy; and the simulation accuracy of multi-temporal before florescence is higher than the simulation results of single temporal. 4, We ran the remote sensing simulation model to simulate Gaoyou rice yields as an example of large scale yield estimation using RS technique. The accuracy reached 94%.