Mountain torrent disaster three-level warning model based on rainfall uncertainty

Mountain torrents are recognized as one of the most harmful natural disasters in the world,which often cause serious casualties,public facilities damage,and environmental disasters because of their sudden occurrence,quick disaster,and great destructive power.In recent years,due to the increase of locally heavy rainfall and the change of underlying surface,frequent flood disasters occur in China.Critical rainfall is an important index of early warning of mountain torrents,and the temporal and spatial distribution of rainfall has an important influence on it.In the process of critical rainfall calculation,the regional design rainfall pattern is often used to generalize the rainfall process,or only one of the temporal and spatial distribution of rainfall is considered,which is lack of simultaneous consideration of both uncertainties.However,for the actual early warning of mountain torrents,the temporal and spatial distribution of each rainfall is different,which leads to the deviation of critical rainfall. To improve the early warning accuracy of mountain torrent disaster,Luanchuan basin was taken as an example.Using the Gini coefficient and the position coefficient of peak rainfall intensity,nine characteristic rainfall patterns,including partial front uniformity and partial concentration were generalized,and the Copula joint probability distribution of total rainfall and peak rainfall intensity was constructed,and the peak relationship under the risk combination probability of 5 return periods of total rainfall and 7 conditional probabilities of peak rainfall intensity was calculated.Five kinds of rainfall spatial distribution scenarios were set up in which the rainfall center of gravity shifts from upstream to downstream.HEC-HMS mode was taken as a rainfall-runoff model,and the critical rainfall under different conditions was determined by trial method.The influence of the uncertainty of temporal and spatial distribution of rainfall on the critical rainfall was discussed.The critical rainfall data set was established,and the three-level critical rainfall early warning model was constructed and verified. The results show that:When the soil water content is less before rainfall,the rainfall center is in the middle reaches,and the rainfall pattern is uniform,the critical rainfall corresponding to the rainfall process in which the position of the rain peak is in front of it is 23.87% larger than that corresponding to the rainfall process wherein the position of the rain peak is in back of it.When the peak is in the back of rainfall,the rainfall process with uniform distribution of rainfall is 6.53% larger than that with the concentrated distribution of rainfall.At the same time,the critical rainfall corresponding to the rainfall process with the rainfall center in the upstream is 31.61%~48.94% higher than that corresponding to the rainfall process with the downstream rainfall center .The conditional probability of the peak rainfall intensity increases when the probability of the total rainfall in the risk combination probability is constant.In the process of verification,the early warning is carried out according to the third-level critical rainfall early warning model of mountain torrents,all four floods can be warned before the flow in the river near the warning object reaches the disaster causing flow. It is shown that the temporal and spatial distribution of rainfall in the basin has a significant impact on the early warning of mountain torrents,and the critical rainfall values under the different temporal and spatial distributions of rainfall are quite different.So the critical rainfall should be properly modified according to the actual rainfall process in the practice of early warning of mountain torrents.The threelevel early warning model considering the uncertainty of rainfall is reasonable and reliable,which can provide technical reference for the early warning of mountain torrents in other small and medium-sized watersheds.

Project Supported:

Clc Number: