利用高分二号影像对城市黑臭水体遥感监测--以南京市主城区为例

Remote Sensing Monitoring of Urban Black and Odorous Water Bodies Using GF-2 Images: Taking the Main Urban Area of Nanjing as an Example

城市河道水质黑臭直接影响人们生存环境和生活质量,准确识别城市黑臭水体对城市生态文明建设具有重大意义。基于GF-2 PMS数据,利用TSUWI(Two-Step Urban Water Index)指数提取研究区水体分布,在此基础上,分别计算影像的黑臭水体归一化比值指数BOI和NDBWI值,采用面向对象的最优特征选择搜索算法SEaTH(Seperability and Thresholds),确定黑臭水体与一般水体的4个分离特征,构成特征空间,利用K-mean算法进行多特征聚类分析,进行黑臭水体的识别与分类。结果表明,南京市研究区范围内共有6条河道15段存在不同程度的黑臭情况,面积约为0.38km^(2),占城区水域(剔除长江南京段面积)面积的1.98%,经实地调查验证,黑臭水体的识别精度为81%。研究表明该方法可以实现对城市河道黑臭水体的监测与评价,为城市水环境治理提供技术支撑。

The black and odorous water quality of urban rivers directly affects people’s living environment and quality of life. It is of great significance to accurately identify urban black and odorous water bodies for the construction of urban ecological civilization. Based on the GF-2 PMS data, this paper uses the TSUWI index to extract the distribution of water bodies in the study area. On this basis, the normalized ratio index BOI and NDBWI values of black and smelly water bodies were calculated respectively, and the object-oriented optimal feature selection search algorithm SEaTH was used to determine the four separation features of black and odorous water bodies and general water body to form the feature space. The k-mean algorithm was used for multi feature clustering analysis to identify and classify the black and odorous water bodies. The results showed that 15 sections of a total of 6 rivers in the Nanjing study area have varying degrees of black and odor, covering an area of about 0.38 square kilometers, accounting for 1.98% of the urban water area(excluding the area of the Nanjing section of the Yangtze River). Through field investigation, the recognition accuracy of black odor water is 81%. The research shows that this method can realize the monitoring and evaluation of black and odorous water bodies in urban rivers, and provide technical support for urban water environment management.