Monitoring Rice Growth By Assimilation Of Remote Sensing And Crop Growth Model

Remote sensing and crop growth models are both important technologies andmeans for crop growth monitoring. But crop parameters quantitative retrieval basedon remote sensing data is commonly semi-empirical or empirical model based on thestatistics, which is lack of mechanism analysis. It can show the spatial variation of theobserved crop, but is lack of the continuity in time. And crop growth models have amore clear mechanism to be explained, which can describe the whole growth anddevelopment of crops. But it is too difficult to obtain all parameters of thosecomplexity models and the models cannot describe the crops’ variation on thedistribution. Therefore, using data assimilation method to combine remote sensingwith crop growth models, can helps to realize the advantages of both technologies sothat we can have a certain machine rational space-time continuously monitoring andevaluation.The study area is in Dongqiao Town, Xiangcheng District, Suzhou City, JiangsuProvince. Rice is selected as experimental subjects. This article analyses the fullpolarization data obtained by synthetic aperture radar (Radarsat-2full polarization)and synchronous rice biochemical measurement data (crop dry biomass): using thetechnology of object-oriented SVM (Support Vector Machine) classification toidentify rice target for rice distribution, with the rice classification accuracy of87.5%;building and using the water cloud model in the HH polarization (the coefficient ofdetermination is0.75306) to retrieve the temporal variations of the rice dry biomassdistribution in study area.An impact factor which is used to describe the level of the coercion inrice-growing, is introduced in WOFOST model (the World Food Study model) whichis after the completion of the parameters adjusted. Using data assimilation meancombines the remote sensing quantitative retrieval (biomass temporal changes) withthe biomass simulation from the model. Detailed speaking, we can use the radar data to constraint WOFOST model by using PSO (Particle Swarm Optimization) algorithmto obtain the distribution of impact factors regionally, so that a regional rice cropgrowth models which is in line with observations can be built and the growth of ricedry biomass can be shown to achieve the regional continuity monitoring on ricegrowth. The use of RS-CGM (Crop Growth Model) assimilation has a good effect andresearch prospects for crop management decisions, and has a very importantsignificance.