用TM影像进行湖泊水色反演研究的人工神经网络模型.PDF

24 2 Vol. 24, No. 2 2003 3 Mar. , 2003 ENVIRONM ENTAL SCIENCE TM 1 1 1 2 2 , , , , ( 11 , 100084; 21 , ) : , BP , T M CODMn. , 25% , , . , , . : ; ; ; ; : P237; X87 : A : 0250-3301( 2003) 02- 04-0073 An Artificial Neural Network Model for Lake Color Inversion Using TM Imagery 1 1 1 2 2 Wan Jianpin , Chen Shen ton , Jia Haifen , Wan Zhishi , Tan U Wa ( 11 Department of Envi- ronmental Science and En ineerin , T sin hua University, Beijin 100084, China; 21 Faculty of Science and T echnolo y , University of Macao, China) Abstract: The technolo y of artificial neural network was used for inversin water quality parameters from T M ima ery data in the paper in order to study w ater quality and eutrophic status of lake. On the basis of satellite synchronous moni- torin experiment, a BP neural network model was constructed, in w hich concentrat ions of SS, CODMn, DO, T-N , T- P and chlo-a were inversed from Landsat T M data and the accuracy of which was ood, the relative error of w hich could be controlled below 25% . Moreover, the reasons of simulatin error, w ays of improvin model and applications of the model w ere also analyzed in detail. T he results of this research told that based on a smal-l scale of satellite synchronous experiment, the model could be applied successfully in invest i ation, analysis and estimation of lake water quality. Keywords:artificial neural network; environmental remote sensin ; water color remote sensin ; inversion; lake , . , , , . , ) ) ) / 0 , . , 1 [ 1~ 3] . , 111 SeaWiFSM ODIS 2001