Research On Winter Wheat Yield Prediction In Northwest China Based On Distributed Hydrological Model And Crop Model

Crop model is one of the main methods wildly used in the yield prediction currently. With the development of the model and computation, it is possible to research on the environmental factors more elaborately by coupling other optimal models, those used to be ignored.The development of agriculture depends considerably on precipitation in northwest China, and the contents of soil water intimidate significantly the growth of crop in the area. The complex and variable terrain has un-measurable effect of second distribution on precipitation, it affects the distribution and transportation of water in soil. The research describes the effect based on DEM in virtue of distributed hydrological model (SWAT). The distributed hydrological model and crop growth model are linked through datum text files to research the prediction of yield. Finally, the effect of the mend on the accuracy of prediction is evaluated. The main research includes following processes.1. The database of meteorology and soil of Shan Xi province is built. Then sub-watersheds are delineated using digital elevation model (DEM) data. The source files are modified and a new project is built to output the contents of soil water in every layers corresponding to those of crop model. The contents of soil water exported from SWAT model are compared and contrasted with precipitation and observed soil humidity.2. The crop model is modified and the contents of soil water of every layer are calculated. The correlative coefficients are calculated between the contents of soil water simulated by two models and soil humidity observed. The results of analysis of correlation show that the correlation ship between the contents of soil water calculated by SWAT model and observed soil humidity is more significant than that between the contents of soil water calculated by crop model and observed soil humidity.3. The crop parameters of the model inputted the contents of soil water calculated by SWAT model or inputted no output of SWAT model are fitted. The results f parameter test show that the parameters of model imported the contents of soil water provided by SWAT are more steady, and the model is more reliable.4. The validation of predicting of winter wheat LAI, dry matter, maturity date and yield shows that the prediction of yield by the crop growth model imported the contents of soil water calculated by SWAT is more accurate than that by the model inputted no output of SWAT.5. The meteorological data and crop parameters are interpolated densely in order to perform prediction in every HRU. The fluctuating of yield year by year is consistent to that of the observed yield. The spatial analysis of prediction shows that distribution of high yield area and low yield area is consistent to that of observed yield.6. The prediction model between biomass, climate productivity and statistic yield in different DVS are built. The test of prediction shows that the correlations are not significant. So the feasible prediction model needs to be built in the research of yield prediction in the future.Experiments and validations proved that the contents of soil water calculated by SWAT is more accurate, and the crop model ARID CROP linked distributed hydrological model SWAT predicts biomass and yield more accurate than itself does.