In this paper, with a view to local geographical environment and terrain factors, we use documents,meteorological data,physical geographic,digital elevation models,as well as logistic model to assess the criticality of flood events of Lanzhou,which is a city in the valley of the Yellow River.The results showed that,during the last one hundred years,the distribution of river channel flood disasters the river channel flood disasters mostly occur on the floodplains of the Yellow River.Compared with the river channel floods,flood passage disasters occur more frequently and on smaller scales.we established predictive formula of river channel flood events using logistic regression, and found the occurrence probability of river channel flood events has a negative correlation with the elevation and distance from the river.This conclusion is in line with traditional experience,and the prediction accuracy rate of this model is as high as 94.4%;besides,the Nagelkerke R2 of the model is 0.870,which closes to 1,and that means the fitting degree of this model is high.On the other hand,we established predictive formula of flood passage flood events using logistic regression model,and found the occurrence probability of river channel flood events has a negative correlation with the elevation and distance from the river. This conclusion is in line with traditional experience,and the prediction accuracy rate of this model is as high as 74.8%.%以黄河河谷型城市兰州市为研究区,基于原生自然地理及地形要素,使用文献、气象、自然地理资料和数字高程模型,同时应用 GIS空间分析方法并结合逻辑斯蒂回归模型,对兰州盆地洪水事件进行了危险性评价.结果表明,近百年来兰州市河道洪水灾害集中在黄河河漫滩上；洪道洪水灾害发生时间比较频繁,与河道洪水灾害相比,灾害的发生范围较小,具有区域性.利用逻辑斯蒂回归建立河道洪水事件预测模型,得出发生河道洪水事件的概率与高程、距黄河距离呈负相关关系,与传统经验符合,并经过模型检验,预测正确率达94.4%,Nagelkerke R2值为0.870,接近1,拟合优度较好.利用逻辑斯蒂回归建立洪道洪水事件预测模型,得出发生洪道洪水事件的概率与高程、距黄河距离也呈负相关关系,与传统经验符合,经过模型检验,预测正确率达74.8%,预测效果较好.
展开▼
