Cotton Science   2005£¬17(5)£º280-284

 

Study on Combinations of Remote Sensing and Cotton Model to Retrieve Initial Inputs and Parameters
ZHAO Yan-xia1£¬2£¬3£¬ QIN Jun4 , ZHOU Xiu-ji2
(1. Chinese Academy of Meteorological Sciences, Beijing 100081, China; 2.School of Physics, Peking University, Beijing 100871, China; 3. State Key Laboratory of Remote Sensing Science, Jointly Sponsored by the Institute of Remote Sensing Applications, Chinese Academy of Sciences, and Beijing Normal University, Beijing 100101, China;4. Research Center for Remote Sensing and GIS, School of Geography and Remote Sensing, Beijing normal university, Beijing 100875£¬China)ª¤ª¤

Abstract: A remote sensing and cotton inversion model has been established on the basis of assimilation theory, assimilating LAI, and taking three optimization methods (SCE,SA,DE), to combine of remote sensing and COSIM cotton model which is relatively developed well. The inversion model can retrieve initial inputs and parameters needed by cotton model. Therefore, the study can resolve the problems of lack of initial inputs when crop model is applied from spot to region. Inversed parameters are sowing date, planting population, amount of nitrogen and irrigation. Simulation test showed that the inversion model established in this paper was correct at great extent. In addition, we have tried to use the established remote sensing-COSIM cotton inversion model with SCE scheme to inverse parameters and predict cotton yield in 11 counties in north part of Xinjiang province.Generally speaking, the results from inversion model including yields and sowing dates, sowing density, nitrogen application amount, and irrigation amount in the application areas are largely consistent with the estimation of real situation. Quantitative comparison by observed and inversed values at two points shows that application results of remote sensing-COSIM cotton inversion model on the regional scale are better. Yield estimated errors of the two points are-5.8% and-5.1%, respectively, which are lower for large area yield prediction. If COSIM model can improve its simulation veracity, all kinds of errors of results will be much lower.
Key words: remote sensing; crop model; parameter inversion