Debris Flow Automatic Extraction And Evaluation Simulation Based On GF-2 Satellite Data

Mudflow has been the geological disaster that caused most casualties,where the local population and economic development could be negatively influenced.However,benefiting from the development of domestic remote control satellite,there is a dependent data source to surveillance geological disaster.Meanwhile,as the computer's data digging and processing has also been developed in recent decade,the remote control satellite technology was innovatively changed,which transferred from low-efficient,manual operation and time consuming to high efficient,high accuracy and automotive.This article uses the domestic remote control satellite data from “Geological Comprehensive Investigation on Northern Coastal Area” project,which basically research Beijing-Fangshan,Mentougou areas.This project built the mudflow surveillance practical system,which could automatically create the mudflow's data,analysing the feature of mudflow's developing,numerical modelling,risk assessment and disaster evaluation.on the base of getting deeper understanding of the Domestic “Gao Fen II” satellite's data,this article combine the mudflow's developing feature and proposes the idea of using the Support Vector Machine(SVM)to automatically extract the skin-layer's loose accumulations after the mudflow.Then combine the DEM data source build the automatically data digging system.Meanwhile,by choosing the areas with relative complete data source foundation.innovatively create a mudflow risk assessment model that could combine those subjective factors.Then,by using the FLO-2D model,the mudflow disaster's coverage under different rainstorm ranking could be accurately forecasted.This article mainly gets below outcomes:1)the finding from remote sensing image of the mudflow's developing process: The mudflow's developing has close relationship with the source area's size,while the loose solid accumulation generally has strong colour differences comparing with the surrounding subjects.By using this colour feature,this article suggests using the SVM to extract the loose skin-layer's accumulation firstly.Then using DEM models to process the spatial three-dimension analysing to get the scale relationship of mudflow.Then combine the area's situation,using value of slope area to screen on different colour zone.Eventually going to the spot to investigate the accuracy of this automatic practical system.2)By reviewing the relative reference,the traditional mudflow's risk assessment methods are mostly experience assessment,which is strongly influenced by those subjective factors.To avoid such shortage,this article suggests using the support efficient DEA-AHP models to rating the survey area's typical mudflow.Based on the assessment structure,such analyse could be actually combine both subjective and objective factors,where avoid the shortage of too objective of using the AHP assessment alone.Meanwhile,fully use the advantage of super-efficient DEA model.3)By processing the FLO-2D numerical modelling on those damage areas of mudflow,trying to forecast the damage scale.This article will also discuss under the different rainy conditions,the mudflow's developing,affected scale and sustaining time's different change.By doing these,this article will give scientific basis to analyse the impact and risk assessment of mudflow.