城市屋面与路面的遥感提取及污染物累积分析

Remote Sensing Extraction Method of Urban Roof and Pavement and Analysis in Pollutant Load

城市面源污染是受纳水体水质恶化的主要原因之一。屋面和路面的污染物累积与冲刷规律差异较大,有效提取该地物信息是开展城市面源污染评估与管理的基础。研究采用基于像元分类法与面向对象法相结合的遥感提取方法,对深圳市高分辨率影像3种典型建筑物密度区域的屋面和路面进行提取试验,同时与目视解译、支持向量机、面向对象法的提取结果对比;基于湿式采样法对典型的屋面和路面样点进行污染物的单位面积累积量采样,结合遥感提取地物信息估算地表污染物累积量。研究表明:(1)该方法提高了屋面和路面提取的目视效果和分类精度,且各研究区域的分类总精度分别为83.82%、93.02%、88.31%;(2)屋面上COD的单位面积累积量为97.32 mg/m^2,而3种建筑物密度区域的路面上COD单位面积累积量分别为196.02、104.50、50.65 mg/m^2;(3)屋面上COD累积总量分别占各研究区不透水地表的30.05%、34.09%、71.74%,路面上COD累积总量分别占各研究区不透水地表的69.95%、65.91%、28.26%。

Urban non-point source pollution is one of the main reasons to cause water quality deterioration. Load and flushing effects of pollutants from roofs and pavements are significantly different,and effective extraction of roof and pavement information from remote sensing images is the basis of urban non-point source evaluation and management. For three regions of different building densities in a Shenzhen＇s high-resolution remote sensing image,the strategy based on the combination of pixel-based classification and object-oriented method was adopted to extract the roof and pavement information,and the results were then compared to those derived from using visual interpretation,pixel-based classification and object-oriented method techniques;pollutant load per unit area in typical samples from roofs and pavements were estimated based on wet sampling method and then used together with the roof and pavement areas calculated from the remote sensing images to estimate accumulative pollutant load. The study showed this method improved the visual effect and classification accuracy of roof and pavement information extraction,and the overall accuracies were 83.82%,93.02% and88.31% for high,medium and low building density regions,respectively. COD load per unit area of roof was 97.32 mg/m^2,and COD loads per unit area were 196.02 mg/m^2,104.50 mg/m^2 and 50.65 mg/m^2 for the pavements in high,medium and low building density regions,respectively. For all the COD loads from impervious areas in high,medium and low building density regions,total roof COD loads accounted for 30.05%,34.09% and 71.74%,respectively,while the percentages of the total pavement COD loads were 69.95%,65.91%,and 28.26%,respectively.