Research On Key Problems Of Urban Surface Water Using Hyperspectral Remote Sensing Data

With the development and progress of society and the intensification of human activities,the water pollution is also aggravated.Remote sensing,as a large-scale,fast and real-time monitoring means,plays an important role in water environment monitoring.Inland surface waters are of various types,background of the surrounding objects is complex and diverse,the optical characteristics of inland case II water are more complex,which have more requirements on the spatial resolution,spectral resolution and signal-to-noise ratio of sensors.High resolution hyperspectral sensors can achieve fine classification and parameter inversion of terrain objects,and it is the one of the development trends of remote sensing in the future.Therefore,based on the high-resolution hyperspectral remote sensing data and taking urban surface water as the research object,this paper focuses on the key issues of surface water monitoring in the city,including surface water extraction,quantitative inversion of intrinsic optical quantities and water color parameters,and the performance requirements of parameter models for spatial resolution,spectral resolution and signal-to-noise ratio,the main content is as follows:(1)The paper analyzed the typical spectral characteristics in urban city.Four water quality indices,NDVI,NDWI,HDWI and IWI,are used to correlate the spectral characteristics.The results show that shadows falling on buildings are the main factors leading to misclassification.Based on the analysis of different types of shadows and spectral characteristics of water bodies,an anti-shadowing water body extraction method is proposed.Taking different area and different types of the Airborne Hyperspectral Remote Sensing Data as experimental data,the results show that ASWEM method can suppress the shadows effectively,the commission error decreased from 40%,30% and 24% to 14.7%,14.6% and 9.66%,respectively.(2)In the inversion of intrinsic optical parameters and water color parameters,in view of the shortcomings of existing inversion models,on the basis of LSBCC-QAA inversion model,the absorption coefficient of particulate matter in the wavelength range of 700-850 nm in the model is calculated.An improved ILSBCC-QAA model is proposed.The correlation between intrinsic optical properties and chlorophyll a and suspended solids concentration was calculated,and the water color sensitive band was obtained,and the inversion model was established.In the inversion experiments of ground measured data in Jiaxing and Baiyangdian,the intrinsic optical quantities inverted by ILSBCC-QAA and the inversion accuracy of water color parameters are both better than the other methods.It shows that the ILSBCC-QAA model is more suitable for chlorophyll a and suspension of urban surface water.(3)Based on the AISA and ASD measured hyperspectral data,the spatial resolution,spectral resolution and signal-to-noise ratio sensitivity of the extraction model and the inversion model are analyzed.The spatial resolution directly affects the extraction accuracy of urban water body.With the decrease of resolution,commission error is first decrease and then increase.When the resolution is 8 m,the commission is reached minimum value of 4.35%,and the influence of spatial resolution on narrow and long river body is more prominent.The spectral resolution and bandwidth of the sensor have a more direct impact on the inversion of water color parameters.The accuracy of the inversion model decreases with the decrease of spectral resolution and the increase of bandwidth.Equivalent noise reflectance has a direct impact on water extraction model and inversion model.With the increase of Equivalent Noise Reflectivity,both of them show a downward trend.Both water extraction and water color inversion models show that the model with single-band is more sensitive to noise.Through this study,new methods for urban surface water extraction and water color parameter inversion using high resolution hyperspectral remote sensing data are provided.The suitability of the extraction and inversion model for spatial resolution,spectral resolution and equivalent noise reflectance is analyzed,which provides a basis and reference for the subsequent development of imaging spectrometer for inland urban surface water environmental monitoring.