泥石流灾害危险性遥感评估模型研究

Hazard assessment of debris flow is an important premise for montoring and forecasting debris flow, and a crucial part in prevention and control work of the debris flow disaster. As far as we know, this kind of disaster is the result of the interaction of plenty of environmental factors and rainfall. At present, some outstanding problems about the hazard assessment of debris flow are how to choose the evaluation indexes in a wide-range and determine their weight and low accuracy of assessment model. In this reseach, debris flow disaster in Sichuan Province is taken as an example to explore and study these questions from the following aspects. The first work is to construct the evaluation indexes of environmental factors such as topography, geomorphology, land-use and vegetation coverage, by using the remote sensing data, which is to foucs on timeliness and scale effect of remote sensing data in environmental factors extraction. The second work is to optimize the interpolation method by plugging the environmental factors from remote sensing because the accuracy problem which is caused by the sparse distribution of meteorological atations. At the same time, the radar rainfall data is used as the assistant data to resolve "has data no rainfall" problem as so to recontruct the high accuracy rainfall factors. At last, the work is to study partition methods of debris flow hazard assessment based on environmental and rainfall factors, and to establish an efficient and highly accurate hazard assessment model based on fuzzy mathematics theory in the different partition, and then, to focus the analysis on the weights value of the different evaluation factors in debris flow hazard assessment model, for debris flow disaster governance and forecasting service.

泥石流灾害危险性评估是开展泥石流监测和预报的重要前提。泥石流灾害是环境因子和降雨综合影响的结果，而大范围评估因子的提取、不同因子权重确定和评估模型的精度是泥石流灾害危险性评估中遇到的突出问题。首先，本研究以多源遥感数据为主提取地形、地貌、土地利用、植被覆盖等环境因子，重点研究环境因子提取中遥感数据的时效性和尺度效应；其次，针对降雨观测站点稀疏、栅格化降雨数据精度不高的问题，引入遥感提取的环境因子，重点研究观测站点降雨数据的插值优化处理方法，同时以雷达降雨量数据为辅，解决"有数据无降雨"的问题，重建区域范围高精度降雨因子；最后，基于环境和降雨因子，研究泥石流灾害危险性评估的分区方法，在不同分区，基于模糊数学理论，建立一个有效且较高精度的泥石流灾害危险性评估模型，并重点分析泥石流灾害危险性评估模型中不同评估因子的权重值，为泥石流灾害的治理和预报服务。

泥石流灾害危险性评估是开展泥石流监测和预报的重要前提。泥石流灾害是环境因子和降雨综合影响的结果，而大范围评估因子的提取、不同因子权重确定和评估模型的精度是泥石流灾害危险性评估中遇到的突出问题。首先，本研究以MODIS和SRTM等多源遥感数据为主提取地形、地貌、土地利用、植被覆盖等环境因子，重点研究环境因子提取中遥感数据的时效性和尺度效应；其次，针对降雨观测站点稀疏、栅格化降雨数据精度不高的问题，引入前面遥感数据提取的环境因子，重点研究观测站点降雨数据的插值优化处理方法，得出结论在四川省区域范围内OK插值方法具有较好的插值效果，同时以雷达降雨量数据为辅，解决“有数据无降雨”的问题，重建区域范围高精度降雨因子；最后，基于环境和降雨因子，利用信息量模型开展了研究泥石流灾害危险性评估的分区，并在不同危险等级分区中，以Logistic回归模型和Bayes判别预报模型，建立一个有效且较高精度的泥石流灾害危险性评估模型，并重点分析泥石流灾害危险性评估模型中不同评估因子的权重值。项目的研究成果可为大范围泥石流灾害的治理和预报服务。