Analysis And Evaluation Of Regional Water Environment Based On Multi-source Remote Sensing Data

A scientific and effective water environment monitoring system is the key to the modernization transformation of water environment governance.In recent years,remote sensing technology has achieved good application results in water environment monitoring tasks,but the relationship between various remote sensing data sources is not yet clear,and there are many types of inversion models available,without forming a recommended standard algorithm under certain conditions.In addition,existing research on inland large scale lakes and small-scale county and district river water bodies is mostly limited,with relatively single water type and water quality indicators,It is of great significance to study the assessment method of overall water environment quality in many types of water areas such as prefecture level cities.For this reason,this study discussed and analyzed the application effect of multi-source remote sensing data in water environment monitoring in Prefecture-level city,and conducted field simultaneous data collection experiments with the reservoirs,artificial lakes,rural rivers,urban rivers and industrial rivers flowing through Fuyang River in Handan City as the study area.On this basis,the relationship between multi-source remote sensing data is studied,and eight algorithms including linear,quadratic polynomial,cubic polynomial,index,support vector,Random forest,extreme gradient lifting,and K-nearest neighbor are introduced to build the inversion model of water quality parameters.Combined with remote sensing images,the water quality in the study area is spatially inverted to obtain the water quality status distribution map,By analyzing the spatial distribution characteristics of water quality and the water quality changes of rivers under different external natural conditions,it is hoped to provide reference and technical support for effective protection and comprehensive management of regional water environment.The main innovative points and conclusions of the thesis include the following:(1)Based on the overall comparative analysis of ground hyperspectral,unmanned aerial vehicle multispectral,and satellite multispectral remote sensing data,it is found that there is a strong correlation between satellite multispectral and unmanned aerial vehicle multispectral.Similarly,the reflectance values of the same wavelength remote sensing at this point are in the order of satellite multispectral>unmanned aerial vehicle multispectral>ground hyperspectral.After a comprehensive study by region,it was found that the correlation between ground hyperspectral data and unmanned aerial vehicle data is strongest in reservoir waters,the correlation between ground hyperspectral data and satellite data is strongest in urban river waters,and the correlation between unmanned aerial vehicle data and satellite data is strongest in urban river waters.After pixel scale classification based on UAV image and comparison with Satellite imagery,it is concluded that near infrared band is greatly affected by pixel scale classification in urban rivers and artificial lake waters,which is of great significance for UAV data and satellite data fusion in this area.(2)After verifying and evaluating the eight constructed models,it was found that refined segmentation of the dataset is beneficial for improving model accuracy.Each model has different applicability in the inversion process of water quality parameters for different water bodies.Among them,the extreme gradient enhancement model has strong generalization ability in water quality monitoring based on ground hyperspectral and unmanned aerial vehicle multispectral data,and overall performs better in unmanned aerial vehicle multispectral data,including turbidity,total nitrogen,total phosphorus,suspended solids The model determination coefficients of ammonia nitrogen and chemical oxygen demand in each water area are all above 0.9,the overall root mean square error is 6.11 ntu and 0.58,0.02,22.78,0.12,4.57 mg/L respectively,and the overall average absolute error is 4.72 ntu and 0.37,0.02,12.46,0.09,3.57 mg/L respectively;In the water quality monitoring model based on satellite data,the random forest model has stronger generalization ability.The model determination coefficients of turbidity,total nitrogen,total phosphorus,suspended solids,and ammonia nitrogen in each water area are all above 0.7.The overall root mean square error is 13.99 ntu and 0.22,0.06,27.59,0.37 mg/L respectively,and the overall average absolute error is 12.5 ntu and 0.17,0.05,22.38,0.29 mg/L respectively.(3)Based on drone images and satellite images,the normalized difference water index and improved normalized difference index were used to extract water bodies from rivers and lakes,and the optimal models selected by the two were combined to invert various water quality parameters to obtain the distribution map of water quality status in the study area.From this,the turbidity of each water body was obtained as follows:industrial area river>rural river>urban river>artificial lake>reservoir;The suspended solids are in the order of industrial zone rivers>artificial lakes>rural rivers>urban rivers>reservoirs;Chemical oxygen demand is in the order of rural river>artificial lake>reservoir>urban river>industrial zone river;The order of ammonia nitrogen is rural rivers>artificial lakes>urban rivers>reservoirs>industrial areas rivers;The total nitrogen is in the order of rural rivers>urban rivers>reservoirs>industrial rivers>artificial lakes;The total phosphorus sequence is rural rivers>artificial lakes>urban rivers>industrial areas rivers>reservoirs.At the same time,based on the statistical analysis of upstream and downstream water quality changes,it is known that after the river enters the reservoir,turbidity,suspended solids,ammonia nitrogen and total nitrogen show a downward trend,while total phosphorus and chemical oxygen demand show an upward trend;After entering the artificial lake,turbidity,total nitrogen,ammonia nitrogen and chemical oxygen demand showed a downward trend,while total phosphorus and suspended solids showed an upward trend;After flowing through the entire rural river channel,all water quality parameters increase;After flowing through the whole urban river,turbidity,total nitrogen and suspended solids increase as a whole,while total phosphorus,ammonia nitrogen and chemical oxygen demand decrease as a whole;The turbidity,total phosphorus and suspended solids increase as a whole,while total nitrogen,ammonia nitrogen and chemical oxygen demand decrease as a whole after flowing through the river of the entire industrial zone.Different water quality parameters vary in the flow of different water bodies and are greatly influenced by the external natural environment.