Generalized Ridge Estimate And Its Application In The Quantitative Evaluation Of The Extent Of Desertification

In recent decades, Generalized Ridge Estimation is a new multivariate analysis method, Rapid development in the theory and applications. The same as the least squares estimation and Ridge Estimation, Generalized Ridge Estimate is mainly used for processing morbid design matrix, it also could solve the problem which is ordinary multiple linear regression can not solve.Based on the above question, taken a typical example in Yijinholo County, inter Mongolia. Based on RS and GIS data. With mathematical methods, through the study of the Desertification impact factor. As can be seen, the two indicators of vegetation coverage and biomass closely related to the degree of desertification. In the desertification quantitative evaluation, Vegetation cover and biomass is the most common and the most important two indicators. Easy to field investigation and obtain the RS and GIS data.On the basis the relationship between vegetation coverage and biomass factors were studied. Least squares estimation and the ridge estimation methods as examples. The results show that there is a serious multicollinearity relationship between the impact factors. Led to the observation matrix show is morbid. So that the square matrix will have one or more close to zero characteristic roots. If solve the parameters to be determined of estimation model by the least square estimation.Will not gets the optimal solution. So choose Ridge Estimation and Generalized Ridge Estimate to replace the least squares estimation. To reduce the adverse impact of solve the undetermined parameters which is caused by multiple linear relationship between the variables. By analyzing Ridge track map, screened out the main factors affecting the estimation of vegetation cover. In order to determine the Generalized Ridge Estimate the value of the parameter, the reference parameter iterative algorithm. And estimation models were established. Respectively establish estimation model, by examining the estimation equation. Generalized Ridge Estimate compared to the ridge estimate, it has a smaller mean square error and higher estimation accuracy. Therefore, it has a high practical value.