Study On Remote Sensing Classification And Change Of Taihu

Extensive development and utilization of water resources caused more serious water shortage and water pollution.The protection of water resources has become an urgent problem to be solved.Long sequence,high frequency,fast and accurate monitoring of water pollution and water quality change is the key to the protection and management of water resources.The traditional monitoring method is to measure the field sampling,according to the measurement results will be divided into different categories of water quality,namely water classification.Obviously,this method has a large workload in the field,and the sampling points are sparse,so it can not meet the needs of monitoring the water quality changes in a large area.This paper introduces the satellite remote sensing technology,using 2003~2011 1711 MERIS images to research the classification method of Taihu water and water type changes in space and time,in order to provide basic support for regional water monitoring and control.Based on the fuzzy clustering of the measured spectral data,combined with the water quality parameters and the optical properties to determine the water type and set up decision tree model.Then,the accuracy and sensitivity of the MERIS synchronization data are tested,and the sixth band is selected as the optimal band.Based on the central band of MERIS and the water quality parameters of synchronization points,which used to adjust the band and adjust the threshold value of the decision tree model,and established a decision tree to classify the water bodies.Finally,9 years of MERIS data in Taihu were used to classify the water body,the spatial distribution characteristics of water bodies in nine years and the changing rules of annual,seasonal and monthly were analyed.The main work and innovation of this paper are as follows:(1)The water classification method based on the measured spectral clustering,combined with the water quality parameters and the inherent optical properties,get water type and the establishment of decision tree classification model based on the measured spectrum of water.(2)The decision tree model based on MERIS image classification is constructed by combining the measured data and the synchronous data,and the applicability of the MERIS image in Taihu water monitoring is verified.Based on the statistical analysis of the 1711 landscape images of Taihu in 2003~2011,the spatial distribution and temporal variation of water bodies in the past nine years were analyzed.Results show: ?1 The chlorophyll dominated by water mainly distributed in Zhushan lake,Meiliang Bay and Gonghu Lake area of the north of the lake;chlorophyll and suspended matter dominated by water mainly in Zhushan lake,Meiliang Bay,Gonghu lake,the north area of eastern coast and the Central Lake area;suspended matter dominated by water mainly in the Central Lake area.?2 In spring and winter,the water area dominated by suspended matter is the largest and the frequency is high;in summer and autumn,chlorophyll and suspended matter dominate the water body area,and the frequency is high;in summer,chlorophyll dominated water was the most obvious.?3 The monthly variation and seasonal variation of the three types of water distribution area and frequency are similar.