Landslide Risk Analysis Based On Multi-Source Data Fusion

Landslide is one of the geological disasters,which could result in serious hazard.Many countries like Japan,India,Italy,USA,China and so on has the annual economic losses amounting to over billions of US dollars due to landslides.Landslide is one of hotspot of international community.Landslide is controlled and influenced by many factors;landslide susceptibility analysis is based on multi-source data,and do the regional landslide risk assessment and carry out effective risk management.It is the main measure of reducing the loss of life and property caused by landslide hazards and great important to scientific and realistic significance.The thesis takes the regional landslide risk analysis theory as the foundation,and adopts data fusion as the main technical measures,to achieve the purpose of typical and different predisposed landslide risk analysis.On the basis of field investigation,relevant data and existing research achievement,start with multi-source data collection and information extraction,then do the research ofthe landslide risk analysis model based on multi-source data fusion and subsystem model,including landslide susceptibility prediction model,the period probability on different predisposed factors,landslide magnitude,landslide hazard analysis,vulnerability of elements evaluation index system,the hazard-affected object value assessmnet,landslide risk model and assessment.Finally,the multi-source data fusion landslide risk analysis is achieved.The main conclusions and results are as follows:(1)Establishing the landslide risk model based on multi-source data,and designing the main feature of the subsystem and the relationship between the each part.(2)Extracting landslide information through multi-source data,including landslide interpretation?disaster environment and inducing factor of landslide.Putting forward the methods of information extraction based on remote sensing data and other data.A method of the landslide and buffer sampling based on the grid evaluation unit is proposed.On this basis,the useful evaluation attribute is extracted by the method of the minimal correlation and the maximal importance degree.A data fusion model of clustering-correlation vector machine landslide susceptibility prediction is established.The results show that the evaluation factors of the Three Gorges Reservoir head region(ZiguiBadong segment)and Lushan earthquake are different,which is mainly related to the relationship earthquake-induced landslides and earthquake-related parameters.Different types of landslides should select different environmental factors.(3)The thesis establishes the time probability analysis method of different evoked factors and landslide magnitude grading system.Due to the occurrence of landslide and induced factors are directly related,so the time prediction is predicted by calculating the probability of occurrence of different predisposing factors.The rainfall induced landslide's time prediction is calculating the rainfall size of different rainfall periods and generating regional rainfall spatial and temporal distribution.The earthquake-induced landslide's time prediction is estimating the probability of annual characteristic earthquakes through historical earthquake data.At the same time,the paper establishes the density classification system of landslide intensity.Then the risk of landslide is calculated by Bayesian network model,which the parameters are the susceptibility?probability of time and landslide intensity.(4)Establishing the landslide risk calculation sub-model,including the vulnerability analysis system,value of the hazard-affected object assessment,risk quantification and risk standard.It is proposed that the vulnerability is constituted by the exposure,sensitivity and adaptability,the assessment of the hazard-affected objectand the risk are divided into the life and economy.(5)The landslide risk analysis is realized on the different stages of water storage in the Three Gorges Reservior head region and 4·20 Yaan earthquake in the Lushan.Then do the comparison between the the different stages of water storage in the process and result.And do the summarizing about the similarities and differences of the different evoked factors on the landslide risk analysis.