遥感分类精度对湿地模拟预测的影响分析

The Analysis of Classification Accuracy of Remote Sensing Data on the Wetland Change Prediction

湿地是生态系统中最为重要的一个生态系统,同时也是近些年来遭受人类活动破坏最为严重的生态系统。近些年来,关于湿地的保护也引起了人们的广泛关注。本文以三江保护区的1999年和2007年两期TM遥感影像为基础,结合人工神经网络和元胞自动机模型,通过对不同分类精度的一系列遥感分类影像作模拟预测,比较分析它们的预测精度,最后结果表明:分类精度和模拟预测精度有着正相关的线性关系,当分类精度达到(Kappa系数)0.75以上时,预测的精度(Kappa系数)才可以达到0.69以上。这一结论为利用元胞自动机模拟预测湿地的空间格局演化时分类精度的选择提供了一定的指导。最后,用此元胞自动机模型预测了2015和2023两年的该区域湿地空间格局图,为三江保护政策的实施提供了一定的依据。

Wetland system is a very important ecosystem.However,it is also destroyed severely by human.Nowadays,the wetland protection has been concerned more and more.In this paper,combined with Artificial Neural Network（ANN） and Cellular Automata model（CA）,two remote sensing images（1999 and 2007） in the Sanjiang Plain were used to analyze the effect of classified accuracy on the prediction accuracy of wetland change.The results show that classification accuracy has a positive correlation with the predict precision.When the classification precision is 0.75（Kappa Index） the predict accuracy is 0.69（Kappa Index）.This conclusion can help us to do predictions using the remote sensing data.At last,two predicted maps（2015,2023） were generated,which can give a certain suggestion to the coming Sanjiang Plain protection policy＇s execution.