Study On The Identification And Classification Of Urban Black And Odorous Water Bodies By Coupling Spectral Characteristics And Black And Odorous Water Quality Indicators

As an important natural element,urban water body is not only the basis for urban ecological environment construction,landscape and biodiversity maintenance,but also the main space for citizens to enjoy water and get close to nature.It also has significant landscape,entertainment and ecological value.With the development of urbanization,urban water pollution is becoming more and more serious.At present,there are many research results on urban water body monitoring by remote sensing,but there are few methods to consider the spectral characteristics of water body and black and odorous water quality index.Based on the measured data and GF-2 image data,this paper analyzes the spectral characteristics of urban black and smelly water bodies and constructs three spectral threshold models to evaluate the accuracy and applicability of different algorithms.Based on the statistical analysis of the measured water quality data,the characteristic bands of water quality parameters were used to construct a random forest water quality inversion model.The accuracy of the model was evaluated by using the root-mean-square error and mean absolute error,and the accuracy of the three algorithms was evaluated by the ground measured points and synchronous image points of DO and ammonia.Black nitrogen（NH₃-N）,total phosphorus（TP）,chemical oxygen demand（COD）and other black and odorous water quality indexes were mapped.On this basis,the coupling identification model of urban black and smelly water quality index was constructed,and the accuracy of the model was evaluated based on the measured data of synchronous image.Finally,the black and smelly water in Guangzhou was identified by remote sensing to reveal the spatial distribution characteristics of black and smelly water in Guangzhou.The main research work of this paper is as follows:（1）Through the spectral analysis of water bodies in Guangzhou,it is found that compared with the general water bodies,the remote sensing reflectance of urban black and smelly water bodies shows a relatively gentle trend in the range of 400-600nm.The spectral characteristics of light and heavy black and smelly water bodies are basically the same,but the average remote sensing reflectance of heavy black and smelly water bodies is lower than that of light black and smelly water bodies.Based on these features,this paper constructs a single band algorithm of green light band,slope index method of black and smelly water and classification algorithm of black and smelly water to identify black and smelly water in Guangzhou.According to the calculation results of sample points,three algorithms are determined to judge the black and smelly water threshold.The accuracy of the three algorithms was evaluated by using ground measured points and synchronous image points.The classification index algorithm of black and smelly water has the highest accuracy,and can separate mild and severe black and smelly water,and the threshold is relatively stable.（2）Analysis of water quality parameters and construction of stochastic forest inversion model.Through statistical analysis of water quality data（correlation analysis,principal component analysis and cluster analysis）,the water quality can be divided into general water body,light black odor water body and heavy black odor water body.Combined with the criteria for judging black and odorous water in the guide and the characteristics of water quality in Guangzhou,the chemical oxygen demand（COD）,dissolved oxygen（DO）,ammonia nitrogen（NH₃-N）,total phosphorus（TP）,etc.are selected as the black and odorous water quality indicators,the characteristic band of black and odorous water quality is selected as the independent variable,and the water quality concentration is the dependent variable.The inversion model of black and odorous indicators is constructed using random forest algorithm,and the root mean square error（RMSE）,mean absolute percentage error（MAPE）Average absolute error（MAE）,determination coefficient R² and other indicators are used to evaluate the model.The MAE of COD is 0.522,MAPE is 10.52%,R²is 0.85,and RMSE is 0.64;The MAE of DO is 0.575,MAPE is 8.2%,R² is 0.87,and RMSE is 0.61;The MAE of TP is 0.503,MAPE is 9.6%,R² is 0.9,and RMSE is 0.58;The MAE of NH₃-N is 0.6,MAPE is 7.8%,R² is 0.84,and RMSE is 0.55.The water quality in Guangzhou is predicted by the constructed model,and the pollution distribution of water quality indicators such as DO,NH₃-N,TP and COD is analyzed.It is found that the concentration of DO is generally high,indicating that the content of DO in water is high,and the water is less polluted by DO;The concentrations of NH₃-N TP and COD are generally high,and the concentrations of NH₃-N,TP and COD in many water bodies deteriorate to Class V,causing serious water pollution.（3）Based on the study of spectral characteristics of water bodies and black and odorous water quality indexes,this paper proposes a classification model for urban black and odorous water bodies,which is coupled with spectral characteristics and black and odorous water quality indexes.Sample points of synchronous images are used to evaluate the accuracy of the model,and the recognition accuracy reaches 93.3%.The spatial distribution of black and smelly water bodies in Guangzhou was analyzed by applying the coupling degree classification identification model to the GF-2 image of Guangzhou in 2019.The overall water quality of Guangzhou is good,but there are still some black and smelly water bodies.The number of black and odorous water bodies in Baiyun District was the largest,with some sections of Shijing River and the upper reaches of Liuxi River being in a slightly black and odorous state.Heavy black and smelly water bodies were found near Wu Chung and parts of the Pearl River Channel in Huangpu District;In Liwan District,there are black and odorous phenomena in Dachongkou Chung,Hesha Chung,A surname Chung,Jiao Kou Chung,Subway B Chung and Subway C Chung,and most of them are bad ClassⅤwater bodies.Some parts of Beihaoyong river in Haizhu District were found to be mildly black and smelly.Heavy black and smelly water bodies mainly distributed in the vicinity of Zhujiang Chung in Huangpu District and part of Liede Chung in Tianhe District.Black and odorous water bodies in Yuexiu District,Zengcheng District,Panyu District and Nansha District are less.On the whole,the quality of water in Guangzhou is good,and the number of black and smelly water bodies decreases,indicating that the control of black and smelly water bodies in Guangzhou has achieved remarkable results.Construction and application of coupling degree model.