滑坡泥石流大尺度统计预报模型的实时检验

利用滑坡敏感性分布和降雨阈值公式建立了一个滑坡泥石流统计模型,该模型可以用于中国大尺度范围内的滑坡泥石流预警。使用CMORPH卫星降水驱动该统计模型,对2016—2017年的106起滑坡泥石流事件进行了验证分析。结果表明,该模型能较好地预警大多数滑坡泥石流事件,其中对72. 1%的雨季滑坡泥石流事件能较好预警,但对非雨季的事件只有35%能较好预警,对雨季的预警效果明显优于非雨季。由于滑坡泥石流主要发生在雨季,因此该模型总体上具有较好的效能。该模型对于强降雨引发的快速滑坡事件具有较好的预警能力,但对于由强度较小、持续时间较长的降雨引发的慢过程滑坡事件的预警效果有待提升。利用该统计模型以及CMORPH实时卫星降水产品,可以建立滑坡泥石流大尺度实时预警系统,对滑坡泥石流减灾防灾具有一定意义。

Real-time warning test of landslide and debris flow with a statistical model in large scale

A statistical model of landslide and debris flow is established by using the landslide susceptibility distribution and the rainfall threshold formula.The landslide and debris flow in large scale across the country can be warned by this model.The statistical model was driven by CMORPH satellite precipitation in this paper,and 116 landslide and debris flow events occurred from 2016 to 2017 were validated and analyzed.The results show that the model can warn most of landslide and debris flow events,and 72.1% of the events in rainy season can be warned,while only 35% of the events in non-rainy season can be forecasted.The effect of warning in rainy season is better than that in non-rainy season.Since the landslide and debris flow events mainly occurred in rainy season,the model is considered to have good performance.In addition,the model has a good warning ability for rapid landslides and debris-flows caused by heavy rainfall,however,the warning effect of the slow landslide events triggered by lower intensity and long duration rainfall needs to be improved.Using the statistical model and CMORPH satellite precipitation real-time products,a real-time forecast system for landslide and debris flow in large scale can be established,which has certain significance for landslide and debris flow reduction and disaster prevention.