The model of Minjiang River woodland expert classification was established by analyzing the rule of the spectral curve which was exeracted from the TM images with the help of the fifth forest resource continuous investigation data in the.

In this paper,the rule of hyperspectral curves of different types of wetland in Dongtinghu Lake were analyzed,a model of decision tree for the final classification was constructed by the Erdas Image expert classification,and vegetation classification supported by GIS data and phonological information was made by using 2-seasonal ETM data of Dongtinghu Lake wetland.

To improve remote sensing image classification accuracy,analysis is made on the mechanism of the traditional unsupervised classification,supervised classification and expert classification.

Based on the climatic and topographic factors,an expert classifier is used to make an agroclimatic regionalization of Beijing Pear.

Based on selected evaluation factors and expert classifier of ERDAS software,reference to national soil erosion grade classification, rapid and precise evaluating the spatial difference of soil erosion could be achieved.

After rational screening of evaluation factors of soil erosion, it conducts quick and accurate evaluation on intensity of soil erosion and spatial difference and verification by using the observed data of hydrometric stations and expert classifier provided by ERADS image processing software and refe.