Studies On Dynamic Prediction Of Rice Yield In County Based On Crop Model And GIS

Crop models have a significant role in helping people to understand the relationship among crops, meteorology, soil and cultivation techniques ,and have become one of the important means of quantitative evaluation in agricultural production. It is one of the important trends in current crop models application how the crop models developed on the field scale will be applied to regional crop growth and yield's simulation and forecast. The successful applications may provide the information of large-scale crop production conditions and development trends for supervisors, and provide decision-making government departments with necessary foundation in their doing agricultural production's macro-control etc..Based on previous reports on crop model construction and regional applications,We further explored crop models implication in regional crops yield prediction, and established dynamic prediction technology of county rice yield based on crop model and GIS, to provide similar study with reference. Specific results are as follows:(1) RCSODS improvementFirstly, the phenology model parameters of RCSODS were simplified using the relationship between temperature sensitivity parameter P and Q and the relationship between P and the basic growth phase coefficient K. There were six parameters in simplified phenology model, including four development stages'basic growth phase (sowing-emergence, emergence-panicle initiation, panicle initiation -heading and heading–maturity) coefficient K, and photosensitive stage's photosensitive coefficient G and critical day length D' .Secondly, RCSODS's photosynthetic production model was improved in the following four aspects, involving revising the original daily assimilation algorithm by using light intension of leaf surface to simulate group photosynthetic production, improving the original temperature impact daily assimilation algorithm by revising the effect of temperature on photosynthetic response curve parameters, daily assimilation amount calculated by hourly method, photosynthetic production parameters simplified being four, including maximum photosynthetic rate, maximum specific leaf weight, photosynthetic efficiency in weak light condition and extinction coefficient.Finally, the improved model was validated using the experimental data of the national 1st agricultural meteorological observation stations in Kunshan Suzhou and Ganyu Lianyungang. The results showed that the RMSE were 2.9d and 1134kg/ha for development stages and yield respectively, and the NRMSE were 1.3% and 13.2%, indicating the improved model may simulate rice growth period well and simulate rice yield relatively well.(2)Scale techniques for application of crop models in countyUsing inverse distance squared to analyze the effect of interpolation station number on the interpolation results,showing interpolation error showed a trend of decreasing at first and then increasing, with the increase of interpolation station number, optimal interpolation station number changed greatly with space, and there was not the best interpolation site number applied in the entire region. However, there was public farthest point weight with relatively higher interpolation precision. When interpolation were used in a region, it was reliable that the weight of farthest site around the interpolation points was controlled in the vicinity of the certain value.Using space integration technology to realize scaling up analysis of the data :crop varieties, soil and cultivation managements. and taking the years 2000-2005's rice yield simulation of counties (districts) in Suzhou and Lianyungang as an example,the technology were tested. The RMSE and NRMSE of simulation value in Suzhou were 742kg/ha and 8.7% respectively, and the RMSE and NRMSE of simulation value in Lianyungang were 975kg/ha and 12.4% respectively, indicating that the technology may realize rice yield forecast in county relatively well.(3) Dynamic prediction of Rice yield in Jiangsu's counties based on crop model and GISDynamic prediction of rice yield for the counties in Jiangsu province were made using improved RCSODS and GIS database having been constructed, combing with the estimation program of rice yield model in county obtained by space aggregation method, and compared with the statistical yields. Results showed that, in addition to some county, simulation errors were generally less than 10%, and the overall prediction effect was better, indicating that the technique could achieve the dynamic prediction of county rice yield well.