Study On Area Change Of Large Lakes In China Based On MODIS Time Series Observation

Lakes are an important part of the ecosystem supporting the survival and reproduction of many species,and has an important influence on human production and life.Changes in lakes can effectively reflect the effects of climate change and human activities over time.In the decades of rapid development of remote sensing technology,scholars have made rich and fruitful progress in the study of lake changes.However,there are still some shortcomings.For example,scholars have not studied the trend of the large lakes in China,and there are still some shortcomings in the identification of water bodies.Based on the shortcomings of the existing studies,this paper discusses the changing trend of ten lakes in China(Poyang Lake,Dongting Lake,Taihu Lake,Hongze Lake,Hulun Lake,Xingkai Lake,Bosten Lake,Qinghai Lake,Nam Co,Selin Co)from the aspect of Lake area.This paper takes MODIS/MOD09Q1 as the data source and uses the long time series image recognition algorithm to extract and analyze the lake area.The main research contents and results of this paper are as follows:(1).A multi-rule system for water body identification.This paper constructs a multi-rule water body identification system,which fully guarantees the accuracy of water body identification under different time and space conditions.There are three rules in the system.First,the dynamic programming threshold algorithm.By using the characteristics of double peaks and valleys in image histogram,the threshold of image segmentation is determined dynamically,which solves the problem of selecting the optimal threshold of image at different times and locations.Second,quality control and cloud detection.We remove low-quality images by visually examining.The cloud and cloud shadows are removed using the data quality description layer sur_refl_state₂50m in the MOD09Q1 data set.Third,human-computer visual interpretation.For remote sensing images which thresholds can not be calculated automatically by the algorithm,we use visual interpretation to extract water area.(2).Long-term variation analysis of large lakes.The article uses MODIS data to extract the monthly area data of China's large lakes from 2000 to 2017 and analysis from interannual and seasonal changes.According to the geographical location,this paper divides ten lakes into three lake areas,the middle and lower reaches of the Yangtze River,the Northeast Lake area and the Western Lake area.It is found that Poyang Lake and Dongting Lake are the most special two of the ten lakes,and they are also the two lakes with the most dramatic changes.The area of two lakes have drastic seasonal interannual variations.Taihu Lake,Chaohu Lake and Nam Co Lake belong to the lakes with little change.Their interannual and seasonal variations are very small.Considering the recognition error,the range of change is reasonable.The other five lakes have unique interannual trends,but one thing in common is that their seasonal trends are very small and can even be ignored.The research provides a solution for water extraction of long time series,contributes to decision support for further lake management and protection,and lays a data foundation for the follow-up study of water level and water reserves.The accuracy of the lake water extracted is fully guaranteed after verification.In addition,the longterm water body identification algorithm used in this paper can be used as a reference for other research.