Inversion Of Water Quality In Nine Plateau Lakes In Yunnan Based On Multi-source Remote Sensing Data

Nine plateau lakes in Yunnan province(hereinafter referred to as the "Nine Lake")play irreplaceable roles in the social and economic development in Yunnan province.With the acceleration of industrialization,the increase of urban population,and the change of people's life style in the Nine Lake Basin,the lake pollution is increasingly serious.Therefore,more and more attention has been paid to the environmental monitoring of nine lakes,one of the most concerned about is the water quality of the lake.This research established the inversion model by remote sensing data source,MODIS satellite images(464 scenes)and HJ-1B images(856 scenes),based on water quality data during 2010-2014 and validated the model by indexes,including the correlation coefficient(R),root mean square error(RMSE)and the efficiency coefficient(NSE),thus the inversion of water quality parameters(chlorophyll a,total nitrogen,total transparency and total algae cell density inversion)of nine lakes were realized.The main results are as follows:(1)With MODIS as the data source,153 groups of chlorophyll a concentration inversion model,163 groups of the total nitrogen inversion model,146 groups of transparency inversion model,and 128 groups of total algal cell density inversion model were established.R2,RMSE,NSE were used to select 33 chlorophyll a concentration inversion model,34 total nitrogen inversion model,35 transparency inversion model and 34 total algal cell density combination model;with HJ-1B image data as the data source,86 groups of chlorophyll a concentration inversion model,77 groups of the total nitrogen inversion model,86 groups of transparency inversion model and 86 groups of total algal cell density inversion model were established,9 chlorophyll a concentration inversion model,7 total nitrogen inversion model,9 transparency inversion model and 9 total algal cell density inversion model were selected.(2)After comparison of water quality retrieval models of nine lakes based on MODIS image and HJ-1B image,it was found that the optimal band combination of the nine lake inversion model of the two data sources did not show obvious regularity and the inversion results were similar,which illustrated the relationship between the statistical model and the spectral information was not obvious.Despite the high resolution,HJ-1B images are affected by the cloud cover and other factors,which make the data continuity too poor to meet the requirements of inversion and to retrieve the nine lake water quality parameters.Therefore,the MODIS data are the main data source for the analysis of the long-term changes of the nine lake water quality parameters in this study.(3)After the inversion optimization models of water quality parameters in nine lakes were applied to MODIS images during 2010-2014,2070 maps of chlorophyll a concentration inversion,1840 maps of total nitrogen inversion,1840 maps of transparency inversion and 1840 maps of total algal cell density inversion were retrieved.After the optimization models were applied to HJ-1B images,1840 maps of the chlorophyll a concentration inversion,1380 maps of total nitrogen inversion,1840 maps of transparency inversion and 1840 maps of total algal cell density inversion were obtained.(4)The spatial distribution maps of water quality parameters of nine lakes at the time near the ground sampling time were chosen to be analyzed.The results showed that the spatial distribution of the water quality parameters in nine lakes is in accord with the professional judgment,which were reasonable.From the perspective of spatial distribution,the water quality of Dianchi Lake,Erhai Lake,Chenghai Lake,Qilu Lake,Xingyun Lake,Yilong Lake,Fuxian Lake and Yangzong Lake decreased from the center to the border.On the contrary,the water quality of Lugu Lake increased from the center to the border.(5)From the perspective of changes during a year,the ground water quality data and remote sensing inversion of water quality showed the consistent trend of change:The maximum values of water quality parameters in the nine lakes are mainly appear from June to July,and the minimum values are between February and March,which shows the tendency to increase first and then decrease at the same time;From the annual variation,the variation of water quality in Fuxian Lake,Lugu Lake,Chenghai is relatively stable,water quality of Dianchi,Yangzonghai,and Erhai shows a tendency to improve,water quality of Yilong Lake,Xingyun Lake improves slightly,water quality of Qilu Lake has a tendency to deteriorate.In a word,the optimal inversion model based on different time periods and different water featres can simulate spatial distribution and dynamic variation characteristics of water quality parameters well.