基于支持向量机的六安市城区多源景观格局演化分析

Evolution of Landscape Pattern Analysis Based on Support Vector Machine of Multi-sources' Satellite Images in Lu'an Urban Area

通过收集六安市城区多源多时相卫星影像数据,采用支持向量机分类方法和GIS技术,分析了2009—2017年的景观格局演化情况。结果表明:(1)Sentinel-2影像地物解译特征显著,图像分辨率满足地物类别识别,总体分类精度达92.41%;(2)在景观结构动态变化中,水域基本保持稳定,建设用地面积增长了约3.4倍,且向东扩展趋势明显;耕地和林地面积总量逐年下降,在东部区域趋于破碎化;未利用地则受建设用地扩大速度的影响,一直处于波动变化中;(3)在景观格局演化中,水域保持动态平衡状态,建设用地和耕地增加量分别达80.61%和55.64%,林地和未利用地面积减少明显,分别达63.39%和72.33%;(4)演化后期的建设用地和林地演化速度明显加快,而耕地面积速度开始下降,城市扩展显著。

By multi-sourcesJ and multi-temporal satellite images data gathering, the evolution of landscape pattern in 2009 - 2017 years in Lu’ an city was analyzed by using support vector machine classification and GIS technology. The results showed： （1）The feature interpreting of image object is significant on Sentinel-2 satellite image, also the image resolution to meet the object classes identified standard and the overall classification precision reaches 92.41%. （2） In the dynamic change of Landscape structure, the water area remained stable, construction area in-creased by about 3.4 times and the eastward expansion tendency is obvious ;the total amount of arable area and forest area decreased year by year in the eastern region and tends to be fragmented; unused areas is affected by the expansion of construction area, which has been in the wave chan-ges. （3）In landscape evolution, water area maintains the dynamic equilibrium state, construction area and arable area increased respectively by 80.61% and 55.64% , forest area and unused land reduced significantly by 63.39% and 72.33% respectively.（4）In the later stage of evolution, the construction area and arable area has been speeded up obviously, and the evolution rate of arable land began to decline, also the urban ex-pansion is remarkable.