Research On The Estabishment Of Landslide Monitoring Database System And The Prediction Model Of Landslide Hazard

In recent years, the landslide monitoring, prediction and assessment based on the development status, distribution, and formation mechanism of landslide has achieved fruitful results. However, the applicability and accuracy of the landslide prediction models are subjected to certain restrictions due to different geographical environment, complex generation of landslide and incomplete landslide data. In this paper, landslide prediction and analysis were made through two different spatial scales(individual scale and regional scale). Meanwhile, the landslide prediction system has been established based on a spatial database and prediction models, which provides technical support and software services for the decision makers. Methods and major achievements in this paper are as follow.(1) The prediction and analysis of individual landslide is divided into spatial and temporal scales. In the landslide prediction of temporal scales, Saito model, nonlinear dynamical chaos sequence and Grey-Markov model are used to predict Huangci landslide, new Wolong temple landslide and Xintan landslide in short-term, middle-term and long-term three different time scales, respectively. And in the landslide prediction of spatial scales, two different perspectives of deterministic and non-deterministic, limit equilibrium and numerical analysis, BP neural network, are used to analysis the stability of Yunnan Simao landslide and loess landslide.(2) First, main factors in the prediction and analysis of regional landslide were extracted, followed by the establishment of the evaluation index system based on the influential factors. Finally, the Fuzzy-AHP model and the information model for 10 intensity zones of Wenchuan are utilized to assess the landslide risk based on the evaluation system. According to the statistical results of validation data, the verification landslide point of Fuzzy-AHP model fall on the areas from very low to very high risk is 0.2%, 8.6%, 25.4%, 28.2% and 37.6%, respectively. And corresponding result of the information model is 3.46%, 5.03%, 7.81%, 35.3% and 48.4%. In addition, from the analysis of the hazard zonation, high risk areas both methods are mainly distributed in the southwest and slightly in the northeast of the study area. Therefore, the use of the two prediction models to predict the regional landslide is effective and accurate.(3) The establishment of the spatial database of landslide monitoring follows the general construction process: conceptual model, logical model and physical model. The landslide attributes database and geographic information database to store landslide data are set up by using the ArcSDE database engine and Oracle database.(4) A prediction and analysis system of landslide was constructed by using the current technology of spatial database and prediction models. It not only provides an information management platform to fast store and manage the spatial data, but also predicts individual and regional landslide. In conclusion, it is a good platform for professional landslide researchers and related workers and an attempt and exploration of prediction and analysis system of landslide.