| The Qinghai-Tibetan Plateau(QTP),the"third pole"of the earth,is the largest alpine ecological region in the world.It is extremely sensitive to current global climate change and is the hot spot for global ecological research.The time series data of Normalized Difference Vegetation Index(NDVI)based on optical satellites can reflect the temporal and dynamic characteristics of surface vegetation,and are the most widely used basic data in ecological remote sensing research.However,limited by the complex terrain and cloudy weather conditions of the QTP,the currently available NDVI time-series data such as coarse spatial resolution(such as 250 m MODIS NDVI)or discontinuous time-series(such as 30 meters Landsat NDVI),still cannot meet the application requirements.For this reason,this paper researched and reconstructed the NDVI dataset with high-spatiotemporal-resolution(30 m-8 days)on the QTP from2000 to 2020,and revealed the high-spatiotemporal variation characteristics of the plateau lake water area based on the dataset.The specific research work include:(1)By fusing the 30-meter Landsat and 250-meter MODIS NDVI time series data,the high-spatiotemporal-resolution NDVI time series(referred to as QTP-NDVI30)was developed on the QTP from 2000 to 2020,with a spatial resolution of 30 meters and a temporal resolution of 8 days.The development of QTP-NDVI30 used all available Landsat 5/7/8 images(>100,000 scenes)in the plateau area from 2000 to 2020,and the reconstruction method was based on the recently developed Gap-Filling and Savitzky-Golay filter(GF-SG)method.Besides,a snow pollution removal module was added to improve the applicability and robustness of the method on the QTP.We conducted a comprehensive accuracy evaluation on QTP-NDVI30.First,we randomly selected 100cloud-free Landsat image areas during the growing season(from July to September)on the QTP.Compared with the reconstructed data,the Mean Absolute Error(MAE)was0.022,and the Spatial Structure Similarity(SSIM)was 0.941.Second,we compared QTP-NDVI30 with the cloud-free 3-meter Planet images in areas with obvious topographical changes on the QTP,and observed consistent spatial changes(average SSIM=0.874).We further estimated the vegetation green-up dates with a spatial resolution of 30 meters based on the QTP-NDVI30 product,and compared with the vegetation green-up dates based on 250m MODIS product.The two have similar spatial distributions in space.But QTP-NDVI30 provides richer spatial detail change information.QTP-NDVI30 has been published in the National Qinghai-Tibet Plateau DataCenter(https://data.tpdc.ac.cn/zh-hans/data/80ee374d-b956-4c51-9572-ee4f6017e0d7),the download volume has reached 3257.52TB(as of March 15,2023).(2)The high-spatiotemporal-resolution features of QTP-NDVI30 will bring new opportunities for the study of plateau ecology.We used QTP-NDVI30 to reveal the spatiotemporal variation characteristics of lake area on the QTP from 2011 to 2018.First,all lake area changes with an area larger than 1 km~2 on the plateau were detected,based on the improved Land Trendr change detection model,and the specific time of lake area change can be detected(Landsat original data cannot obtain the change time).We randomly selected 100 lake change objects and compared the detection results of QTP-NDVI30 with 3m Planet images in August,and found that the two have high consistency in time and space(Overall accuracy is 0.701,precision is 0.878,and recall is 0.807).In addition,we analyzed the changing trend and main driving factors of lake changes on the QTP.The experimental results showed that the change characteristics of lakes on the QTP from 2011 to 2018 are mainly divided into three stages:the first stage(2011-2015)is a slow growth period,and the lake area generally expanded,but the expansion rate decreased(Trend=-24.97 km~2/yr);the second stage(2015-2017)is a period of rapid growth,and the total area of lakes increased rapidly(Trend=469.22km~2/yr);the third stage(2017-2018)returned to a period of slow growth,and the decline rate of lake expansion in this stage(Trend=-133.74 km~2/yr)is greater than the first stage.Second,we analyzed the change of lakes with different sizes.The results showed that the lake area was within a certain range(<50km~2),and the change area of the lake was positively correlated with its own area.Finally,we studied the driving effects of climate characteristics(potential evapotranspiration,temperature and precipitation in different seasons)on lake changes of QTP.The results showed that precipitation in spring and winter(From October to April)are the main driving factors for lake changes on the QTP,with a contribution of 53.4%,followed by the summer and autumn(from May to September)precipitation with a contribution of 16.14%and the temperature with a contribution of 13.42%. |
PDF Full Text Request