Research On Remote Sensing Interpretation Algorithm Of Black And Odorous Water Bodies Based On Fuzzy Decision Tree

In recent years,under the efficient implementation of the "Ten Water Regulations" promulgated by the State Council,black and smelly water bodies have received extensive attention and active discussions from experts in the field of water environment,combined with convenient and efficient cutting-edge remote sensing technology,which effectively compensates for the traditional black and smelly water monitoring in time,Lack of human and financial resources.Through the long-term and timely monitoring of satellite images,experts can grasp the current distribution of water bodies in the study area,and effectively monitor the obvious black and odorous water bodies in a certain period of time.These developments have provided strong technical support for the national prevention,supervision and evaluation of black and odorous water bodies in recent years.As black and odorous water bodies are mostly small bodies of water and scattered and difficult to find,the problems of black and odorous water bodies in counties are revealed one by one,bringing the focus of attention and treatment direction of black and odorous water bodies closer to the public's vision.In this dissertation,93 test pits and ponds were designed and deployed in Langfang City,Hebei Province,and remote sensing identification of black and odorous water bodies was carried out for operational monitoring.Different from the predecessor's limitation of black and smelly water monitoring to a single threshold method,this dissertation is based on the Sentinel-2A satellite image in the two-scene study area acquired in August 2018,and uses an improved fuzzy decision tree to synthesize multiple water feature information to achieve black and smelly water,which realized the construction of the identification model of Langfang City.The main research results obtained in this dissertation are:(1)According to the quantitative and qualitative methods,the field data of the study area was discriminated,and the difference in optical characteristics of the two types of water bodies was deeply analyzed.It was found that clean water bodies and black and odorous water bodies were distinguishable in terms of suspended solids concentration and chlorophyll a concentration.The concentration of suspended matter is above 110mg/L and the proportion of black and odorous water is about 70%.The clean water body in the area with the concentration of Chl-a below 43mg/L occupies more than 85% of the total,and the overall distribution of clean water is vertical,concentrated at a concentration of 0 Within the range of ?45mg/L,the black and smelly water is concentrated in the area where the concentration of chlorophyll a islower than 38mg/L(2)By comparing the reflectance characteristics of clean water and black and odorous water in the visible light band,it is found that the reflectance of black and odorous water is generally gentle in the blue to red wavelength range,roughly below the clean water body,and the clean water exists in the green band.Obvious spikes,the two types of water bodies are indeed separable in the visible light band.The threshold method for distinguishing black and odorous water bodies is reliable and feasible..(3)Fuzzy decision tree model has obvious advantages over single feature threshold segmentation Based on the implementation process of the fuzzy ID3 algorithm,through the K-means clustering method and the fuzzy information gain rate to expand the classification attributes,the control of the decision tree growth mechanism is realized,and the classification accuracy of the uncertain and continuous data sets is improved.Depending on the interpretation,the final black and smelly water recognition model has a correct discrimination rate of 76.0%.