| Sri Lanka is an important participant in the"21st Century Maritime Silk Road".As a hub of Asia-Africa-Europe sea routes,it has important economic and geographical significance.The domestic rainfall is abundant but the spatial and temporal distribution is unbalanced,there is obvious seasonal water shortage,and it is affected by floods for a long time.It is located in the tropical region with high temperature and large evaporation capacity,which causes the satellite transit image affected by clouds and fog.Meanwhile,the water environment is complex,and there are many factors affecting the extraction of water,which increases the difficulty of obtaining accurate water information.Therefore,studying the rapid extraction of surface water and flood information can provide scientific basis for water protection,management and planning.It is also important practical significance for flood disaster assessment,rescue and resource allocation.Based on the Google Earth Engine and multi-source remote sensing images,this paper makes a comprehensive analysis of surface water and flood in Sri Lanka.First of all,in order to extract water accurately in complex water environment and accurately estimate the area of each kind of surface water in Sri Lanka,the LBV transformation equation of Sentinel-2 image is derived.Based on the analysis of the characteristics of water and other typical features after LBV and K-T transformation,a water extraction model based on LBV and K-T transformation is proposed.The extraction results of single-band threshold method,OTSU based water index method(NDWI,MNDWI),classifier model(SVM)and the model presented in this paper are compared from visual interpretation and quantitative analysis.At the same time,based on OTSU threshold method,Sentinel-1 image was used to extract water,and compared with the results of the Sentinel-2 image.The water extraction model proposed in this paper was used to extract water information of Sri Lanka in 2019,and the detailed information of water in the whole island was calculated and its distribution characteristics were studied.Then,based on the Google Earth Engine and Landsat-5/8 images,the water rapid extraction model was established.The spatiotemporal change characteristics of surface water in 1995,2005 and 2015 were analyzed.Finally,based on Google Earth Engine and multi-source remote sensing data,a rapid extraction model of flood inundation range was established,which was applied to the flood research in Sri Lanka in May 2017.At the same time,the rapid extraction platform of flood range was established and published.The main research results are as follows:(1)Compared with single-band threshold method,OTSU based water index method(NDWI,MNDWI)and classifier method(SVM),the water extraction model based on LBV and K-T transformation has the highest overall accuracy,reaching 98.13%,which effectively eliminates the influence of cloud,cloud shadow,mountain shadow and cement field,and achieves the accurate extraction of water in complex water environment.The model can be widely used in regions of cloudy,mountainous or complex water environment.(2)In 2019,the water area of Sri Lanka was 2015.61 km~2,accounting for 3.07%of the total island area,which includes many types of water bodies.It has a large number of small lakes and reservoirs,and the overall situation is reduced from northwest to southeast.The larger lakes and reservoirs are mainly concentrated in the north-central region.The total area of the lakes and reservoirs water was 1226.20km~2,which reaches 60.84%of the national water area.Rivers are mainly concentrated in the north-central of the country,and their length is generally short,with an area of 228.57km~2,accounting for 11.34%of the national water area.Lagoons are mainly distributed in the coastal regions of Sri Lanka,and the quantity is large,with an area of 560.84 km~2,accounting for 27.82%of the national water area.(3)The water extraction model based on Landsat-5/8 image and Google Earth Engine can be quickly applied to the study of temporal and spatial changes of long-time surface water.From 1995 to 2015,the water area of Sri Lanka increased significantly,with an increase area of 483.66km~2.The area of the lakes and reservoirs increased significantly,followed by the lagoon,and the river area increased slightly,basically unchanged.In the past 20 years,the quantity and area of the four levels of the lakes increased in different amplitude.The lower the level of the lakes,the greater the increase of the quantity,and the area of the IV level lakes increased significantly,increasing by 197.49km~2during the past 20 years.The increase of the number and area of lakes and reservoirs was mainly concentrated in arid provinces such as northern province,northern central province and eastern province.(4)By applying the fast extraction model of flood range based on multi-source remote sensing image and Google Earth Engine to the flood inundation analysis of the special flood in Sri Lanka on May 27,2017,it showed that the model can be applied to the flood inundation analysis quickly and accurately.The constructed flood submergence rapid extraction platform can be widely used in flood research in Sri Lanka. |
PDF Full Text Request