基于MOD13A2数据的长三角区域土地覆盖类型变化研究

Study on the change in land cover of Yangtze River Delta based on MOD13A2 data

本研究利用MOD13A2产品的NDVI数据,构建决策树分类方法,获得长三角地区2003—2013年的土地利用/土地覆盖（LULC）分类图,在此基础上分析了长三角地区近11年LULC的时空变化特征。结果表明：通过分析各LULC类型的年变化特征构建的决策树分类模型,能较高精度完成研究区的LULC分类,基于2003年Landsat ETM＋数据分类结果验证,总分类精度达86.09%,Kappa系数为0.75;2003—2013年长三角地区耕地、林地、水体和建设用地面积占比分别在47.08%~54.2%、28.9%~32.45%、4.82%~4.91%和11.64%~15.67%之间变化,其中,林地面积年平均增长1.35%,建设用地年平均增长3.35%,耕地面积年平均减少1.45%,水体面积变化较小;各LULC类型面积的年变化率与统计年鉴数据相比,林地高估,其他LULC类型相差不大,这也进一步揭示该分类方法在长三角区域进行LULC变化监测具有可行性。

In this paper,the Land-Use and Land-Cover （LULC）in Yangtze River Delta from 2003 to 2013 were examined by building the classification decision tree and using the NDVI data of MOD13A2.The results indicated that the Landsat ETM＋data in 2003 was used to evaluate the accuracy of the classification methods.The overall classification accuracy of the decision tree was 86.09%,and Kappa coefficient was 0.75.The areas of farmland,forest land,water,and construction land changed in 47.08%-54.2%,28.9%-32.45%,4.82%-4.91%,and 11.64%-15.67%,respectively,from 2003 to 2013.The growth rates of forest land area and construction land area were 1.35 and 3.35%/a,respectively,while farmland area decreased markedly and water area was almost similar.By comparing the classification image and statistics data,it was found that forest land was over-valued and the other LULC types were similar,which proved that the classification method used for LULC monitoring is feasible.