Study On The Effect Evaluation Of Cold Damage On Rice Yield Based On MODIS And TM

China is a major country in paddy rice production. The national economy and the people's livelihood are greatly subject to stability of rice yield. In recent years, the frequent occurrence of cold damage brought about significantly negative effect on China's paddy rice production, and subsequently led to great loss in rice yield, which seriously threatened China's food safety. At present, the effect of cold damage on paddy rice has been a hot issue in related study field. Research involved dominantly focuses on the establishment of crop mechanism models with so many agronomic parameters which are collected by filed experiments. But the crop models method has its own disadvantage that core parameters (e.g. LAI) are very difficult to be available and its applicable spatial scope is limited so that the calculated rice yield per unit acquired by the method could not be representative at larger spatial scales. The thesis took Wuchang city of Heilongjiang province characterized by frequent cold damage attack as study area; applied remote sensing technology featured by timely, large-area and low-cost data acquisition to calculate the daily LAI required by the SIMRIW model; and calculated rice yield per unit for the year 2006 for the study area under the condition of cold damage effect.An TM imagery of May 31^st and a time series of 23 MODIS imageries which dates relatively evenly distributed from the transplanting to the maturity of the rice growth were selected to be a basis for the study. After pre-processing including cloud testing, atmospheric correction, projection transformation and band fusion, the visible interpretation was applied to extract rice sown areas so as to create vector-based layer. The 23 NDVI layers derived from the results that the 23 MODIS imageries were dissected by the vector-based layer and from thereafter band calculation. The 23 mean NDVI values of NDVI layers were calculated for LAI retrieving. In the process of LAI retrieving from NDVI, linear model, logarithm model, parabola model, 3 order polynomial model and exponential model were applied to study the relationship between LAI and NDVI respectively. The error evaluations of 5 statistical models by using two indices of R² and RMSE showed that the computation result of parabola model was the most suitable one to fit daily LAI values. Two kinds of fitting methods which were aggregation fitting and section fitting were applied in the daily LAI acquisition from the transplanting to the maturity. The aggregation fitting of the daily LAI values was wholly realized by 3 order polynomial model; meanwhile, section fitting of the daily LAI values was partially finished by exponential model, logarithm model and linear model according to the change characteristics of LAI value in the rice growth phase. After error analysis of two fitting methods, the output of the aggregation fitting method was adopted to continue calculation on account of being superior to the section fitting method. In terms of the requirement of SIMRIW model, the LAI values from the emergence to the transplanting were calculated by the linear model so as to obtain the daily LAI values for the whole rice growth stage.Input of the daily LAI values for the whole rice growth stage to SIMRIW model with the calculation of agronomic parameters could make it possible to work out the average rice yield per unit, 5411.74kg hm^-2, in the study area which accounted for 83% of the actual value. And comparison analysis of the results of the rice yield per unit by using SIMRIM in different countries was complemented and the effect extent of cold damage on rice yield per unit was also analyzed. The thesis analyzed the error between calculated rice yield per unit and actual one and discussed the error reasons from the different angles of NDVI calculation, LAI retrieving, LAI fitting, SIMRIW model parameters calculation and application and LAI calculation statistical models application.The main conclusions of the thesis are as follows: (1) the core parameters(LAI)of crop mechanism model acquired by remote sensing technology could improve the accuracy of the calculation of paddy rice per unit under the condition of cold damage effect. (2) the calculation result of paddy rice per unit based on the combination of remote sensing technology and crop mechanism model could be more representative at regional level; (3) the daily LAI values could be retrieved and fitted by using statistical models, meanwhile the accuracy was acceptable but need to be improved further; (4) the SIMRIW model could be reasonably simplified to calculate rice per unit with only a set of agronomic parameters under the condition of rice variety with similar physiological and phenological characteristics and similar cropping.The innovations of the thesis are as follows: (1) application of remote sensing technology in crop mechanism model for the effect of cold damage on rice yield per unit improved the efficiency of data acquisition and the calculation accuracy and extend applicable scope of crop mechanism model. (2) the thesis first fit the daily LAI values according to the change characteristics of LAI in the rice growth phase. (3) upscaling of the calculation result of paddy rice per unit from field scale to regional scale was realized.