Water Quality Remote Sensing Inversion And Spatiotemporal Analysis On International Lake

Lake Xingkai is the the largest international lake of China and looks like a splendid sapphire inlaid on the Northeast China.In 1996,as the international waters located between the two countries,the Chinese and Russian governments signed the"Xingkai Lake Protected Area Agreement",which included the two national-level protected areas in the Xingkai Lake basin as internationally important wetlands and world biosphere reserves.In recent years,rapid population growth and sociometric development have promoted anthropogenic utilization on Lake Xingkai.The discharge of industrial and household wastewater,increased irrigation caused by a large amount of cultivated farmland,water withdrawal from farmland,pollution from pesticides and fertilizers,and large amounts of water and soil erosion after deforestation,etc.,have damaged the self-rehabilitation ability and aquatic ecosystem of Lake Xingkai,leading to water quality deteriorates and harmful algal blooms occurred frequently.As materials related to water quality monitoring of Lake Xingkai are difficult to obtain and there are few international investigations on the spatiotemporal variation of water quality in Lake Xingkai,using satellite images to evaluate the water quality parameters of Lake Xingkai has become the most effective and realistic research method.Based on 301 measured samples obtained from 29 field surveys conducted in Lake Xingkai from 2012 to 2018,this paper constructs remote sensing estimation models of water quality parameters applied for Landsat series sensors and MODIS remote sensing satellite images.And lacustrine eutrophication index of Lake Xingkai was calculated via the comprehensive trophic state index method.The spatiotemporal evolution pattern of several water quality parameters and the trophic state index of Lake Xingkai got successfully analyzed.Furthermore,the interannual changes of water quality parameters in the waters of China and Russia in Lake Xingkai are taken as the research objects,and the driving factors of anthropogenic and natural factors in China and Russia are analyzed.The main conclusions are as follows:(1)The evaluation of atmospheric correction of satellite product images.In this paper,107 water surface reflectance data measured in five field experiments from 2017to 2018 were selected,and the blue,green,red,and near-infrared bands corresponding to Landsat and MODIS images were used to construct a regression model.The results showed that the Landsat remote sensing reflectance products had a good efffect of atmospheric correction.The blue band with the lowest correlation with the measured data has reached 0.74 and the slope of the regression line is very close to the 1:1 line and most of the samples fall at 95%prediction interval.Although the atmospheric correlation of MODIS product data is generally not as good as Landsat,the green band with the lowest correlation has also reached 0.6,the regression line is very close to 1:1,and almost all the samples fall in the 95%prediction interval.Therefore,the reflectance products of the Landsat series sensors and the daily reflectance products MOD09GA/MYD09GA of MODIS have good atmospheric correction effects and could be used for long-term sequence inversion.(2)The weight assigned of comprehensive eutrophication index method for Lake Xingkai.According to the measured samples of all valid water quality parameters obtained from 2012 to 2018,the linear regression of the Chla concentration with SDD,COD_Mn,TN,and TP was performed,and the regression R²between the Chla concentration and other four indicators were calculated.The weight value of comprehensive eutrophication index method corresponds to each water quality parameter were calculated and then applied to the TLI calculation of Lake Xingkai.(3)The model construction of remote sensing estimating water quality parameters of Lake Xingkai.This paper successfully constructed remote sensing estimation models of Chla concentration,SDD,SPM,TN,and TP based on 301 samples,which were both suitable for MODIS and Landsat series product data.The study found that the determination coefficients of the ratios of the blue and green bands of Landsat and MODIS to estimate the concentration of Chla in Xingkai Lake were 0.677 and 0.833,respectively,and the R² of the verification reached 0.589 and 0.798,respectively.The determination coefficients of the ratios of the green and red bands of Landsat and MODIS for the SDD of Lake Xingkai reached 0.833 and 0.808,respectively,and the R² of the model verification of the two sensors both was 0.804.The R² of the ratios of the sum of blue,red and near-infrared band to the green band of Landsat and MODIS to estimate the TN concentration of Lake Xingkai were 0.771 and 0.811,respectively,and the verification R² of the two sensors respectively reached 0.625 and 0.794.The R²of the ratios of green band to the sum of blue and near-infrared band of Landsat and MODIS to estimate the TP concentration of Lake Xingkai were 0.805 and 0.802,respectively,and the verification R² of the two sensors respectively reached 0.812 and0.802.Moreover,this paper compares the R² of COD_Mn with Chla concentration and SPM concentration,and on the premise of fully comparing the inversion effect of Chla concentration and SPM concentration,SPM concentration is selected as the indirect inversion index of COD_Mn.An estimation model of COD_Mn based on SPM inversion was constructed,and the R² of the model reached at 0.815.(4)Spatiotemporal variation analysis of water quality parameters in Lake Xingkai.The remote sensing estimation model of water quality parameters based on Landsat and MODIS image data was applied to the filtered 198 scenes of Landsat images and 2212scenes of MODIS daily reflectance products MOD09GA and MYD09GA images,which realized the long-term inversion of water quality parameters in Lake Xingkai.Then the every corresponding TLI raster image was calculated through the TLI index method.The annual average spatial distribution and interannual evolution curves of various water quality parameters and TLI were plotted in units of years.The results showed that since 1984,Chla concentration,SDD,CODMn,TP concentration,and TLI have increased with fluctuation.SPM and TN concentration showed a fluctuated declining trend.Based on the vector boundaries of Lake Daxingkai,Xiaoxingkai,Domestic Xingkai,and Russian Xingkai,the inversion results of various water quality parameters were extracted and calculated.And the water quality indicators of Lake Daxingkai,Xiaoxingkai,Domestic Xingkai,and Russian Xingkai were compared and analyzed.The inter-annual changes and inter-monthly distribution differences show that the differences of all water quality indicators varied among waters,but the overall fluctuation trend are consistent,and the overall inter-monthly distribution differences are not significant.(5)Spatiotemporal distribution of harmful algal blooms in Lake Xingkai.Based on AFAI harmful algal bloom extraction algorithm,198 scenes Landsat images and2212 scenes MODIS daily MOD09GA/MYD09GA products images were processed and analyzed.Results showed that the total frequency of harmful algal blooms observed by Landsat and MODIS images in Lake Daxingkai and Lake Xiaoxingkai were 138 and23,respectively.Mapping the spatial distribution of the most serious harmful algal bloom in each year found that harmful algal blooms in Russian waters are more serious than in Chinese waters of Lake Daxingkai.From the spatial distribution of algae crossing frequency in Lake Daxingkai and Xiaoxingkai,we can know that the algal bloom frequency of Lake Daxingkai is higher than that of Xiaoxingkai,and the algal bloom frequency of the waters in Russia is also higher than that of our country.(6)Driving forces analysis of water quality variation in Lake Xingkai.Based on the single-factor spearman analysis method and random forest regression method,this paper explored the driving forces of natural,anthropogenic factors in the Chinese basin and Russian basin on the eutrophication and water quality variation of Lake Xingkai.The single-factor human impact analysis results showed that the indicators of fertilizer application,gross national product,population increase,total water consumption,agricultural water consumption,and domestic wastewater discharge in the Chinese basin are significantly positively correlated with the TLI in the Chinese waters of Lake Xingkai.Industrial water consumption,domestic water consumption,industrial wastewater discharge are significantly negatively related to the TLI in the Chinese waters of Lake Xingkai.The amount of fertilizer and the gross national product in the Russian basin were significantly positively related to the TLI in the Russian waters of Lake Xingkai,while the population increasement was siginificantly negatively correlated.The single factor natural factor analysis results showed that the factors that significant level lower than 0.1 and positively related with TLI of Chinese waters of Lake Xingkai were summer temperature,summer pressure,and autumn temperature,and negatively correlated were precipitation in summer,autumn and winter.Besides above factors,the factors that significant level lower than 0.1 and positively related with TLI of Russian waters of Lake Xingkai was summer wind speed.The results of random forest regression showed that the contribution rate of the anthropogenic factors in the Chinese basin and the Russian basin that affected Lake Xingkai was significantly higher than the natural factors,and the agricultural water use and domestic wastewater discharge contribution rate in the Chinese basin is the highest,followed by the national production total value,population growth and fertilizer application.However,due to the limited availability of data in the Russian basin,the analysis results show that the contribution rate of GDP and population growth is the highest,and the contribution rate of fertilizer application was lower.