Spatio-temporal Patterns And Drivin Force Analysis Of Water Clarity In Chinese Lakes Using Sentinel-2 Imagery

Lakes,seen as special inland waters on the earth surface,are important natural resources closely related to human survival and development.There are many lakes widely distributed in China.These lakes can be divided into five limnetic region according to geographical locations,i.e.,Eastern Plain Limnetic Region(ELR),Northeast Plain Limnetic Region(NLR),Inner Mongolia-Xinjiang Plateau Limnetic Region(MXR),Tibet-Qinghai Plateau Limnetic Region(TQR)and Yun-Gui Plateau Limnetic Region(YGR),respectively.With the rapid growth of population and urbanization,lake ecosystems face increased environmental pressure,as well as the global warming.Many environmental issues as lake eutrophication,decreased transparency and algal blooms occurred,and massive growth of algal particles and other aquatic organisms also change the underwater light field,which resulting in the degraded ecosystem service function of lakes.Therefore,it is important to strengthen the monitoring frequency and ability for lakes which could provide valuable suggestions and references for environmental policy-makers.This study developed a remotly sensed SDD empirical model using Sentinel-2MSI imagery and their atmospherically corrected normalized leaving-water reflectance.This model was used to map spatial and temporal transparency in investigated Chinese lakes with surface area larger than>1 km~2in 2020.Further,the driving factors of spatial SDD patterns in five limnetic were explored.The main findings of this study were as follows:(1)We collected 431 water samples from 63 field lakes in 2017-2019,and the lake transparency showed significant seasonal differences.The lake transparency increased from spring to autumn,and decreased in winter.In terms of geographical distributions,lake transparency gradually decreased from west to east,with higher transparency were found in the TQR and lowest values in the NLR.In these four study areas(no sampling sites in the YGR),the distribution of lake transparency showed a decreased order as TQR>MXR>ELR>NLR.(2)Based on the correlation analysis between 431 water quality parameter data and transparency,the main reasons affecting the transparency of lakes in the NLR are water turbidity and non pigment particles;Human activities and nutrient input from lakes are the main factors affecting the transparency changes of lakes in the ELR;The non-point source pollution and climate dryness caused by activities such as industry and agriculture are the main factors causing changes in the transparency of lakes in the MXR;The conductivity of lakes in the TQR is high,and the concentration of dissolved ions in the water is high.The input of nutrients causes changes in the structure of phytoplankton in the lakes,further affecting the nutritional level of the lakes.(3)The normalized leaving-water reflectance corrected by C2RCC processor from Sentinel-2 MSI imagery were used to develop a SDD empirical model.The SDD model exhibited a good performance with regressional determined coefficient was0.83 for calibration dataset(N=287,RMSE=0.46 m,MAE=0.36 m)and 0.72 for validated dataset(N=144,RMSE=0.62 m,MAE=0.49 m).Therefore,our model can be used to map SDD distributions in a lage scale.(4)In 2020,considering the different SDD levels as 0-0.50m,0.50-1.00m,1.00-2.00m and 2.00-4.00m,the proportions of lake numbers were 44.33%,17.09%,16.02%and 18.24%respectively.The investigated lakes(area>1 km~2)equipped with>4.00m transparency accounted for 0.04%.We also found that the spatial transparency distributions had a increased order as TQR>YGR>MXR>ELR>NLR.(5)The transparency of five typical lakes in 2015 and 2020 were examined,and the lake transparency showed the ranges of 0-8.00 m.Of which,Qinghai Lake generally showed the highest transparency,and Xingkai Lake has the lowest transparency.Significant interannual variations of transparency for typical lakes from Bosten Lake,Xingkai Lake and Hongze Lake were found.In 2015 and 2020,the transparency of Bosten Lake and Erhai Lake increased.(6)Based on the SDD results,we also examined the trophic state index(TSI)of these lakes in 2020 according to Carlson trophic state index.Considering the different TSI levels as 30-50,50-60,60-70 and>70,the proportions of lake numbers were22.47%,16.02%,17.09%and 44.33%,respectively.Only 2 lakes were counted as poor trophication(0<TSI<30).We also found that the TSI had a decreased order as NLR>ELR>MXR>YGR>TQR.(7)The drivring factors,e.g.,hydrological factors(lake area,lake volume and average lake depth),meteorological factors(average annual precipitation,average annual wind speed and average annual temperature)and socioeconomic factors(spatial distribution of population,gross domestic product)were used to analyze the relationships between them and lake transparency.The results showed that,among the investigated lake hydrological factors,lake volume significantly correlated to lake transparency.Considering the natural factors,the altitude showed a significant positive correlation with lake transparency,signifying the high altitude lakes had high transparency.In addition,among the socioeconomic factors,the population and GDP did not show a strong correlation with the transparency of the lake.In summary,Sentinel-2 MSI images combined with empirical models can achieve large-scale regional transparency inversion,and the results are of reference value.The transparency of lakes in China shows obvious seasonal differences,and the transparency of lakes gradually decreases from west to east.Lake volume and elevation are the main reasons for the differences in transparency.About 80%of lakes in China are in different degrees of eutrophication,and the problem of water ecology and environment should not be underestimated.