A Research Of Yielding Winter Wheat By Using The Method Of Assimilating Domestic Meteorological Satellite Data And Crop Growth Model

Timely crop yield estimation is of significance for national food security and economic development. Therefore, improving crop yield estimation with higher accuracy has been one of hot topics in the field of agricultural remote sensing in the world. This thesis presented the progress on the study of winter wheat yield estimation in the great plain of Hebei which adopted the satellite data assimilation and crop modeling methodology to combine FY-3 MERSI data and WOFOST crop growth model.The major study points are as follows:(1) The winter wheat growing area in the great plain of Hebei in the growing season of 2013 was mapped with the multiple dates MERSI imageries at 250 m resolution by using hierarchical classification method. In comparison with the re-sampled data of the global land cover, the validation showed that the overall classification accuracy was 86.4% and Kappa coefficient was 0.8109. Thereafter, an experiential NDVI-LAI exponent equation was formed based on the in situ observations of the winter wheat spectra data, LAI measurements in the agro-meteorological experiment station of Gucheng, CMA as well as the simultaneous Satellite data. The LAI maps in the growing season were further retrieved from the MERSI data with the abovementioned equation.(2) EFAST global sensitivity analysis method was applied to tune the WOFOST crop model by using the biophysical data and field management data in the winter wheat growing season in the agro-meteorological experiment station of Gucheng, CMA. After the sensitivity test of the WOFOST, the emergence date IDEM, initial dry matter weight TDWI and initial soil efficient water content WAV among the crop parameters and the soil management parameters were chose to be as the assimilation parameters. To meet the model input requirements, meteorological data, crop parameters, and soil and management parameters were interpolated by using the Kriging interpolation method.(3) SG filtering algorithm was applied to get the smoothed MERSI-LAI profile and MODIS-LAI profile. LAI was set as the variable to reach the minimum cost function of tuning WOFOST crop model. Optimization algorithm SCE was used on pix-by-pixel to adjust assimilation parameters including emergence date, the initial dry weight, and initial soil available water content. When the LAI objective function meets the initial setting, the outputted yield information of winter wheat of the WOFOST is the estimation. Referring to the statistics data, the result showed that the yield estimation accuracy of assimilating MODIS-LAI data was lower than the estimation without the assimilation, and RMSE increased by 513.86kg/ha; the yield estimation accuracy of assimilating MERSI-LAI data rose greatly in comparisons with that of the assimilating MODIS-LAI and the estimation without the assimilation, RMSE were reduced by 750.20kg/ha and 236.34kg/ha, respectively. The yield estimation accuracy of assimilating MERSI-LAI data was much closer to the statistics.