长白山地区滑坡灾害的时空危险性分析

发布时间：2018-01-07 04:08

本文关键词：长白山地区滑坡灾害的时空危险性分析　出处：《延边大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着长白山地区近几十年来地质条件的不断变迁、人类在该地区的建设活动日益增多,由此造成的山地滑坡危害逐渐被人们所认知;故对该地区的滑坡灾害从空间和时间上进行预测已成为一个急需解决的问题。地质灾害危险区的评价是深入认识受灾情况的基础,同时也是制定灾害防治规划的科学依据。论文首先对滑坡的影响因子的权重进行单独分析,然后用RES和安全熵法对研究区进行整体性分析,采用GIS对研究区的总体的滑坡灾害危险性进行评价,最后依据危险程度进行分类并对各滑坡分区进行评价。绘制出滑坡灾害危险性分布图。得出极危险区的面积占总研究区域面积的9.9%,较危险区占22.3%,中等安全区占30.2%,较安全区占24.0%,安全区占13.6%的结论。查阅大量文献后,我们发现绝大多数滑坡灾害与降水密切相关。在与水资源有关的学科领域中,有相当多的现象符合Markov性。同时,从理论上看:在诸多的定量分析方法中,Markov过程预测理论不仅具有准确性高、对历史数据量要求较低、依赖程度少等优点,还能使预测效果有很大程度的提高。再从实际情况看:降水诱发滑坡往往具有突发性,常常给人类造成猝不及防的危害。所以本文在搜集大量的历史降水诱发滑坡灾害资料(因诸多现实因素,本论文对延边州的降水诱发滑坡进行分析)的基础上,用Markov过程对研究区的降水状态及其各种降水状态下所诱发的降水滑坡重现周期进行了深入的探讨,绝大多数降水诱发滑坡灾害发生在较强降水年,其次为常降水年,之后是强降水年。最后用实际降水滑坡灾害对得出结论的合理性进行验证,验证结果令人满意。值得一提的是,由于长白山地区绝大部分属于山地地貌,地质条件恶劣,极易在降水时发生山体滑坡;本研究对延边州的降水诱发滑坡预测方法也可应用到研究区的其他地区,进行适当的分析后仍可得出合适的结论。从Markov过程分析的结果看,降水量的预测是对降水诱发滑坡预测的基础性工作,其准确性直接影响到对当年降水状态和降水诱发滑坡的评价。所以,我们采用BP神经网络预测模型对研究区的年降水量进行模拟。当我们将预测数据与真实值进行对比,发现其准确率普遍达到85%左右,印证了该方法对年降水量预测的可行性。在前面我们得出的成果基础上,假如我们预测出某一年很有可能发生由降水所导致的滑坡灾害,那么我们还可以与本论文的滑坡灾害危险性分布图相结合,找出那些滑坡发生危险程度高的地区进行灾害重点防御,从而可减少灾害损失。实现了滑坡危险性评价在空间和时间上的统一。将两者结合起来我们很容易知道在某个时间段在某个特定的区域会发生滑坡灾害,这种时空危险性分析相结合的方法在某种程度上可以克服单一角度分析的局限性(如仅从空间角度或仅从时间角度来分析)并能提高滑坡灾害预报的精度。综上所述,本论文不仅对研究区的滑坡灾害危险性分布进行了研究、还对该区域因降水而诱发的滑坡进行了分析并给出了比较符合该地区的降水量预测模型。为该地区滑坡灾害的"防灾减灾"给出了一定的参考建议。
[Abstract]:With the continuous change of the geological conditions in Changbai Mountain area in recent years, in the area of the construction of human activities is increasing, which caused the landslide hazard gradually recognized; so the landslide disaster in the region from the space and time to become an urgent problem to predict geological disaster risk evaluation. Is the basis of in-depth understanding of the disaster situation, but also to develop the scientific basis for disaster prevention and control planning. Firstly, to carry out the analysis of the weight factor of landslide, and then in the study area were analyzed by RES and the whole safety entropy method, the landslide hazard overall GIS in the study area were evaluated, at last according to the degree of risk classification and the landslide zoning evaluation. Draw the distribution map of landslide hazard. The extremely dangerous area of the total study area 9.9%, A dangerous area is 22.3%, medium security area accounted for 30.2%, the relatively safe area accounted for 24%, accounting for 13.6% of the security zone. Conclusion a large number of documents, we found that the vast majority of landslide disasters and precipitation are closely related. In water related fields, there are quite a few phenomena with Markov. At the same time, from the theory look: in many of the quantitative analysis method, Markov prediction theory not only has high accuracy, low requirement on the amount of historical data, rely on the advantages of the degree, but also make the prediction effect is greatly improved. From the actual situation: rainfall induced landslides are often unexpected, often causing harm to be caught off guard human. So this paper collecting historical data of precipitation induced landslides (due to many practical factors, the precipitation induced landslides in Yanbian prefecture were analyzed on the basis of the study area) by Markov process The precipitation state and various precipitation precipitation state landslide induced by the return period are discussed, the vast majority of rainfall induced landslides occurred in the strong precipitation, followed by precipitation often, after heavy rainfall years. The rationality of the landslide disaster finally concluded on the actual precipitation verification verification with satisfactory results it is worth mentioning that, because most of the Changbai Mountain area belongs to the mountainous topography, geological conditions, prone to landslides occurred in precipitation; the research on Yanbian's rainfall induced landslide prediction method can also be applied to other areas of the study area, the appropriate analysis can still draw appropriate conclusions from the Markov process. The analysis results show that the precipitation forecast is based on the prediction of precipitation induced landslide, its accuracy directly affects the state of precipitation and precipitation induced landslide The evaluation. So, we use BP neural network prediction model of the study area of annual precipitation is simulated. When we forecast data and actual values were compared, found that the accuracy is generally reached about 85%, confirms the feasibility of this method to forecast the annual precipitation. The basis of the results of that I have in front of. If we predict a likely landslide caused by rainfall, so we also can be combined with the landslide hazard map of this paper, find out the high risk areas of heavy disaster defense landslide, which can reduce disaster losses. The landslide hazard assessment unified in time and space. We combine the two is easy to know that in a time of landslide in a particular region, this method of risk analysis of the combination of space-time The limitation to a certain extent can overcome the single point of analysis (e.g., only from the perspective of space or only from the time perspective) and can improve the accuracy of landslide forecast. In summary, this thesis not only landslide hazard distribution in the study area were studied also in this area because of precipitation induced landslide the analysis and the comparison with the precipitation in this region. The region prediction model of landslide disaster prevention and mitigation "gives some suggestions.

【学位授予单位】：延边大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P642.22

【相似文献】