Rainfall-induced Landslides Extraction And Spatio-temporal Characteristics Analysis Based On Multi-source Remote Sensing Data

Landslides are one of the most common and serious natural disasters in the world,posing a serious threat to the safety of people's lives and property.There are many main factors that induce landslides,such as rainfall,earthquakes,human activities,etc.Among them,rainfall-induced landslides(rain-induced landslides)account for about90% of the total number of landslides.Studying the spatio-temporal characteristics of rainfall-induced landslide events is useful for regional landslide warning and Disaster prevention is of great significance.At this stage,landslide data is relatively scarce in most areas,and it is difficult to support relevant research.Extracting landslide spatial information based on remote sensing change detection is an effective means to supplement landslide data.However,the current landslide extraction method is difficult to apply to regional landslide detection,and the extracted landslide type is not clear and the time scale is relatively coarse.Therefore,the research is oriented to regional scale and long time scale.Time-series landslide spatial information extraction methods,rainfall-induced landslide discrimination methods and time identification methods are particularly important for enriching the landslide database and analyzing the spatiotemporal characteristics of regional landslides.In view of this,this research deeply analyzes the advantages and disadvantages of remote sensing change detection methods,and designs a set of GEE-based remote sensing landslide spatial information automatic extraction algorithm;Considers the early soil moisture and recent rainfall of rainfall-induced landslides,and establishes a hydro-meteorological basis.The threshold landslide prediction model realizes the discrimination and time identification of rainfall-induced landslides;Finally,based on the extraction results of rainfall-induced landslides,the regional spatio-temporal characteristics of rainfall-induced landslides are explored.The main research contents and results of this paper are as follows:(1)Extraction of spatial information of landslides based on multi-source remote sensing data: Based on the long-term remote sensing landslide change detection method,taking into account the number and area accuracy of landslide extraction,the NDVIbased dual-temporal landslide change detection method is optimized to achieve regional scale Quickly extraction of spatial information of landslides,while improving the accuracy of remote sensing landslides extraction;(2)Rainfall-induced landslides time identification based on hydro-meteorological threshold: Considering the type and time of landslides,a hydro-meteorological threshold model that takes into account the antecedent soil moisture and recent rainfall is proposed to realize the discrimination and time identification of rainfall-induced landslides.Analysis of spatio-temporal characteristics of rainfall-induced landslide events provides data support;(3)Analysis of the spatio-temporal characteristics of rainfall-induced landslides in different regions: Based on the extraction results of rainfall-induced landslides in the Emilia-Romagna region of Italy and the Dujiangyan region of Sichuan,China,the distribution and evolution of landslides in different regions are studied,and the effects of rainfall-induced landslides are analyzed.The characteristics of spatio-temporal changes can provide an early warning basis for regional disaster monitoring mechanisms.