Landuse/Cover Change Analysis And Prediction Research In The Haidian District Of Beijing

The research based on Beijing Haidian district remote sensing image of Landsat7,including 1999,2006,2013.Using object-oriented method for classification,landscape pattern index and Logistic-CA-Markov model,makes a comprehensive analysis of land use/land cover change(LUCC)in Haidian district nearly 20 years.At the same time,predict its future tendency of Landuse in 2020 by using this model.With the help of Remote Sensing,GIS,simulation model and landscape index,analyses the trend of landuse change.Details as follows:(1)Using Ecognition software built landuse classification criterion aim at haidian district.At first using ESP to determine the scale of segmentation,and executing classification based on Level 29 and Level 10 respectively.then comprehensive consider index such as LWI,SLAVI and NDVI,spectral features and the actual landuse situation of Haidian district,divided it into five classes,including farmland,forest,unused land,water area and building area.verify the classification meets the accuracy requirement for further analysis.(2)In 1999-2013,landuse change of Haidian district presents farmland,forest,unused land and water area decrease,at the same time building area increase.Analyzing on the transfer matrix,we could draw a conclusion that others class mainly transferred into building area and main transferred out class are unuse land and framland.Integrated index,change rate and the degree of landuse change in Haidian district are bigger than zero and rising year by year,it shows Haidian district in high-speed development stage.Edge Density(ED)are increase after first decreased while Splitting index(SPLIT),Shannon’s diversity index(SHDI)and Shannon’s evenness index(SHEI)shows the tendency of slow decline.It proved that along with the social development,advantage landscape in Haidian district began to expand.then the total length of edge between patches increased,so as to make the boundary density increased and the landscape diversity and evenness drop.(3)Using Logistic-CA-Markov model combined with the Gradient,DEM,Population density,Administrative center distance,Main road distance,Railway distance makes a landuse simulation of 2013.Compared with the simulation and classification results in 2013,we found kappa coefficient is 0.84.the reliability of the model was validated.Based on this model and the classification of 2013,We could predict 2020 landuse situation through suitability atlas and Markov transition matrix.Final prediction results show that the landuse change in haidian district from 2013 to2020 will keep the trend of 1999-2013.