EN中文

Journal of Geo-information Science >

2017 , Vol. 19 >Issue 7: 941 - 949

DOI: https://doi.org/10.3724/SP.J.1047.2017.00941

Orginal Article

Establishment and Validation of a Meteorological Warning Model for Landslide Hazards in Sichuan Province

  • LI Yunjun , 1 ,
  • LIU Zhihong , 1, * ,
  • LV Yuanyang 2 ,
  • LIU Jinbao 1 ,
  • WANG Ping 3
Expand
  • 1. Chengdu University of Information Technology, ChengDu 610225, China
  • 2. HUAYUN ShineTek, Beijing 100081, China
  • 3. Guangan Meteorological Bureau, Guangan 638000, China
*Corresponding author: LIU Zhihong, E-mail:

Received date: 2016-08-10

  Request revised date: 2017-04-26

  Online published: 2017-07-10

Copyright

《地球信息科学学报》编辑部 所有
Fold

Abstract

Landslide disaster is serious in Sichuan province. This influence is more obvious after the year 2008. How to prevent landslide disaster is an effective way to reduce landslide disaster losses and protect people's lives and property. Research on early warning models of landslide hazard is the core issue of landslide disaster prevention. This study collected the landslide data, precipitation data between 2008 and 2013, digital elevation data, geological lithology data and seismic intensity data. Our research can be divided into the following two parts: (1) Evaluation of landslide hazard in Sichuan Province. This study used the method of deterministic coefficient to quantify the slope, relief amplitude, hydrogeological lithology vegetation coverage, seismic intensity and annual rainfall factor. We also established a logistic regression model to quantitatively analyze the risk of landslide disaster in Sichuan Province. The results were also verified. The results indicated that the high risk area of landslide disaster in Sichuan is similar to the shape of the letter "Y". The risk value of landslide disaster is as high as 0.97. In addition, the risk of landslide disaster in northeastern Sichuan is very high, with a maximum of 0.8. According to the statistical analysis of the frequency of landslide and the analysis of risk zoning area, the area of region where the value of landslide hazard is between 0.1 and 0.2 accounted for 22% of the whole province's area. The area of the risk value exceeding 0.9 occupies only 5% of the area of the whole province. 35% of the historical disaster points are located in this area, indicating that the degree of landslide disaster risk is high. The spatial distribution characteristics of landslide hazards in Sichuan Province is as follows: the landslide is zonally distributed along the Longmenshan fault zone, the Xianshui River fault zone and the Anning River fault zone, and clustered in the northeastern Sichuan, which is consistent with the model results.(2) Research on early warning model of the meteorological risk of landslide disaster. Based on the statistical analysis of the early landslide disaster and rainfall, and the risk assessment of landslide disaster, this study took the landslide risk assessment as the static factor and the daily rainfall data as the dynamic factor to determine the precipitation probability value, the zoning value of the landslide disaster risk, precipitation probability value of one day ahead, precipitation probability value of two days ahead as the influence factor of the model. The influence degree of each factor to the warning results is decreasing in the order above. Finally, we established the meteorological coupling warning model of the landslide hazards. According to the verification of 2139 disaster points, 80% of the landslide disaster can be successfully predicted, among which 30% of the landslide disaster warning values are more than 0.75. 18% warning values of the landslide disaster are higher than 0.99. 90% of the large and super large landslide disasters can be successfully predicted. 40% of large-scale landslide disaster warning results are greater than 0.75. 12% of large-scale landslide disaster warning results are more than 0.99. On July 10th, 2003, there was a case of group-occurring landslide. It shows that the warning area decreased greatly. Empty quote rate and missing quote rates are greatly reduced compared with the current model results of Sichuan province.

Key words： Sichuan Province; landslide; logistics; warning models; weather warning

Cite this article

LI Yunjun , LIU Zhihong , LV Yuanyang , LIU Jinbao , WANG Ping . Establishment and Validation of a Meteorological Warning Model for Landslide Hazards in Sichuan Province[J]. Journal of Geo-information Science, 2017 , 19(7) : 941 -949 . DOI: 10.3724/SP.J.1047.2017.00941

1 引言

基于气象因素的地质灾害区域预警方法主要分为隐式统计预警方法、显示统计预警方法和动力预警方法3大类[1]。隐式统计预警方法以降雨为单一触发机制,核心内容是对降雨阈值的确定,不能表达地质环境对滑坡灾害的影响,对于地质环境复杂的地区,该方法不能满足实际需求。动力预警方法是物理机制的预警模型,主要依据降雨前、降雨过程中和降雨后降水入渗到斜坡体内的动力转化机制,具体描述在整个过程中斜坡体内地下水动力作用变化与斜坡体状态及其稳定性的对应关系。该方法尚局限于试验场地或单个滑坡区的研究探索阶段。显示统计预警方法将地质环境因素与降雨参数等迭加建立预警模型,充分考虑了预警区域的地质环境因素,比较适用于地质环境模式比较复杂的区域。
四川省早期的滑坡灾害预警模型主要为隐式统计预警模型,仅考虑降水因素,以降水阈值作为滑坡灾害预警的唯一标准进行小尺度预警预报;随后,采用专家打分的方法,对以往的隐式预警模型进行了改进,虽然模型预警精度有所提高,但该方法存在主观性较强,精度较低等缺点,仍不能满足业务要求。
虽然以往四川省滑坡灾害预警模型的研究取得了一定的成果,但受滑坡灾害灾情数据不完整,地质数据、降水数据精度差的限制,缺乏针对全省范围、地质-气象因素耦合、更客观、精细化的地质灾害预警模型[2-4]。
为提高滑坡灾害危险性评价精细化程度和更客观、合理的确定降雨量与滑坡灾害发生概率,本文收集了2008-2013年滑坡灾害灾情数据、30 m分辨率DEM数据、包含加密雨量站和区域气象站在内的小时降水数据等资料,基于目前应用最广泛、适用性最强的显式统计滑坡灾害预警模型[1],建立了报预警模型,以期进一步提高预警精度,限制空报率和漏报率,从而提高滑坡灾害气象预警准确率。

2 数据与方法

2.1 数据源

本文选取国内外研究中普遍适用[5-11]的地形因子、地质因子和气象因子作为静态评价指标,以滑坡灾害发生当日降水量和前期累积降水量作为动态指标,结合四川省环境特征,使用坡度、地形起伏度、地质岩性、地震烈度、植被覆盖指数、年均降雨量、日降水量等因子建立显式滑坡灾害预警模型。主要数据如表1所示。根据国土资源部的《县（市）地质灾害调查与区划基本要求》实施细则,地质灾害预警最佳单元为1~3 km,因此,每个影响因子数据分辨率经ArcGIS软件重采样为1 km。
Tab.1 Model data collection

表1 模型数据整理

数据类型 名称 时间 来源 分辨率/比例尺
文本数据 四川省滑坡灾害灾情数据 2008-2013 四川省地质环境监测总站 9354个
四川省降水数据 2008-2013 四川省气象台 小时
矢量数据 四川省水文地质岩性数据 1981 四川省环境科学研究院 1:20万（53 m）
栅格数据 四川省DEM 2009 地理空间数据云（http://www.gscloud.cn/） 30 m
四川省NDVI 2000-2011 地理空间数据云（http://www.gscloud.cn/） 250 m
四川省地震烈度 2010 四川省地质环境监测总站 3000 m

2.2 研究方法

显式统计预警方法充分反映了预警地区地质环境要素的变化,可以通过不同的耦合手段,灵活地调节地质因子和降水因子的权重系数,随着调查研究精度的提高,能相应地提高地质灾害的空间预警精度[12]。
逻辑回归模型[13]是半定量的多元统计分析方法,用于解决不连续变量的特殊对数线性模型,国内相关研究表明逻辑回归模型在地质灾害研究中具有假设简单,限制条件少、操作简单、能自动筛选影响因子等明显优势,并且解决了二元模型未考虑的地质灾害因子相互依赖的问题[14-15],对同一地区的地质灾害危险性评价或预警所需计算量比支持向量机、人工神经网络方法小,结果准确率较其他方法更高。本文选取逻辑回归模型作为显式地质灾害预警模型中地质灾害危险性评价和地质——气象因子耦合的基本方法。
逻辑回归模型建模方法如下：
Z = ln P 1 - P = A + B 1 X 1 + B 2 X 2 + … + B n X n （1）
P = EXP ( Z ) 1 + EXP ( Z ) （2）
式中：P为地质灾害发生概率,取值在[0,1]之间;（1-P）为地质灾害不发生的概率;X1,…,Xn是影响地质灾害发生的因子;B1,…,Bn是各影响因子对应的逻辑回归系数。
在解决地质灾害问题中,逻辑回归模型受异质数据同化方法的限制[16],分别选取确定性系数（CF）方法[17-18]和降雨量概率化的方法解决地质灾害危险性因子和日降雨量因子异质数据同化的问题。

3 滑坡灾害危险性评价

滑坡灾害危险性评价是在查清滑坡灾害活动历史、形成条件、变化规律与发展趋势的基础上,确定滑坡灾害发生的潜在可能性大小,是显式滑坡灾害预警的基础。

3.1 滑坡灾害危险性区划

在采用二元回归模型进行滑坡灾害的危险性评价时,需要解决2个问题：①滑坡影响因子的选择。本文选定了坡度、地形起伏度、植被覆盖度、年均降雨量、地质岩性和地震烈度6个影响因子。②滑坡因子的量化问题,即异质数据类型的合并问题。本文采用确定性系数（CF）法作为滑坡因子量化的方法。
将剔除误差数据后的8761条有效滑坡灾害历史灾害记录分为建模数据（占总记录的70%）和检验数据（占总记录的30%）。根据张锡涛等[19]、常鸣等[20]的研究,地质灾害点影响范围在2 km内,建立地质灾害点3 km缓冲区,使用ArcGIS软件空间分析功能、矢量数据工具集等功能在缓冲区外随机生成与有效地质灾害点相应数量的未发生地质灾害点。
使用ArcGIS软件栅格数据提取工具提取地质灾害点和未发生地质灾害点致灾因子CF值（图1-6）,以地质灾害是否发生为因变量,坡度、地形起伏度、地质岩性、植被覆盖度、年均降水量和地震烈度CF值为自变量在SPSS软件中建立逻辑回归模型。
Fig. 1 Relationship between slope CF and distribution of landslide hazard points

图1 坡度CF与滑坡灾害点分布的关系

Fig. 2 Relationship between lithology CF and distribution of landslide hazard points

图2 地质岩性CF与滑坡灾害点分布的关系

Fig. 3 Relationship between vegetation coverage CF and distribution of landslide hazard points

图3 植被覆盖度CF与滑坡灾害点分布的关系

Fig. 4 Relationship between relief amplitude CF and distribution of landslide hazard points

图4 地形起伏度CF与滑坡灾害点分布的关系

Fig. 5 Relationship between annual rainfall CF and distribution of landslide hazard points

图5 年均降雨量CF与滑坡灾害点分布的关系

Fig. 6 Relationship between seismic intensity CF and distribution of landslide hazard points

图6 地震烈度CF与滑坡灾害点分布的关系

逻辑回归模型通过对因子相关性的分析判断因子间相互依赖的程度,分析发现,坡度与地形起伏度相关性超过0.5（表2）,二者相互依赖性较高,应合并为一个因子,故本研究只保留坡度因子进行建模。
Tab. 2 Correlation of different factors

表2 因子间相关性

地震烈度 地质岩性 坡度 地形起伏度 年均降雨量 植被覆盖度
地震烈度 1.000 -0.001 -0.008 -0.011 -0.281 0.014
地质岩性 -0.001 1.000 0.011 0.032 -0.152 -0.016
坡度 -0.008 0.011 1.000 0.610 0.015 0.097
地形起
伏度		 -0.011 0.032 0.610 1.000 0.012 0.083
年均降
雨量		 -0.281 -0.152 0.015 0.012 1.000 -0.310
植被覆
盖度		 0.014 -0.016 0.097 0.083 -0.310 1.000
多次随机生成未发生地质灾害点并循环建模,似然比检验-2LL值无明显变化,卡方检验结果较小,说明模型结果稳定;Cox & Snell R Square超过0.4,Nagelkerke R Square超过0.5,表明模型总体拟合度较好;各因子均通过显著水平为0.05的Wald检验,且因子间相关系数很小,相互独立;模型对地质灾害发生的预测准确率达到78.9%。
最终确定地质灾害危险性评价模型如下：
H = EXP - 0.343 X 1 + 1.774 X 2 + 0.742 X 3 + 1.216 X 4 + 1.458 X 5 - 0.144 1 + EXP - 0.343 X 1 + 1.774 X 2 + 0.742 X 3 + 1.216 X 4 + 1.458 X 5 - 0.144 （3）
式中：H为地质灾害危险性值;X1为地质岩性CF值;X2为年均降水量CF值;X3为植被覆盖度CF值;X4为地震烈度CF值;X5为坡度CF值。
通过ARCGIS空间分析功能获得地质灾害危险性区划（图7）,以0.1为间隔,将四川省滑坡灾害危险性分为10级,从分类后的结果可以明显看出,四川省滑坡灾害危险性超过0.8的主体区域呈“Y”字形分布在四川盆地西缘和盆地西北部龙门山断裂带附近、川西凉山州中部鲜水河断裂带附近以及雅安市西部和东部,川东北巴中、南充、达州市也是滑坡灾害高危险区集中分布的区域;滑坡灾害危险性0.4-0.8之间的区域主要分布在除四川盆地外的川东大部分区域,川东南泸州市、宜宾市大部分区域,川西凉山州中部、攀枝花北部,甘孜州鲜水河断裂带附近,川西高山高原区也有零星分布;川西大部分区域、川北若尔盖湿地保护区、四川盆地地区滑坡灾害危险性小于0.4,不易发生地质灾害。
Fig. 7 Risk division of landslide geological hazard

图7 滑坡地质灾害危险性区划

3.2 评价结果的合理性验证

如图8所示四川省滑坡灾害危险性在0.1-0.2之间的区域的面积最大,占全省面积的22%,随滑坡灾害危险性增加面积呈减小趋势,危险性在0.4以上各区间面积波动很小,占全省面积5%左右。灾害点危险性在大于0.9区间达到峰值,占总灾害数的35%,随滑坡灾害危险性降低,灾害点数量迅速减少,危险性在0.4以下的历史灾害很少,只占总灾害数的6%左右。危险性大于0.8的区域仅占四川省面积的20%,但71%的灾害点都集中在这一区域,这一特点与实际情况相符;滑坡灾害危险性大于0.9的区域仅占四川省面积的5%,但35%的历史灾害点分布于这个区域内。
Fig. 8 Proportion of different landslide hazard partition

图8 滑坡灾害分区面积及检验区灾害点危险性比例

同时以ROC曲线检验滑坡灾害危险性评价模型对滑坡灾害发生的敏感性,可得到曲线下面积达0.819,同样说明该模型对滑坡灾害是否发生敏感性很高,能有效提高滑坡灾害预警准确性。
从四川省滑坡灾害空间分布来看,主要有2个显著特征：①条带状分布的特点,这与四川省地震烈度分布相呼应,沿龙门山断裂带、安宁河断裂带和龙泉山脉呈条带状分布的特点尤其明显;②川东北地区滑坡灾害有群发性特征,体现为川东北地区滑坡灾害密度远高于其他地区。如图9所示,滑坡灾害规模级别空间上呈条带状分布和川东北集中性分布的特征明显,特大型滑坡灾害主要分布在龙门山断裂带、零星分布在安宁河断裂带附近,集中分布在川东北巴中、南充、达州3市。
Fig. 9 Distribution of different landslide hazard scales

图9 滑坡灾害规模级别分布

4 滑坡灾害气象风险预警模型建立

在滑坡灾害危险性区划的基础上,以降雨量作为诱发因子建立地质—气象耦合的预警模型,研究发现降雨是滑坡灾害诱发的最主要因素,但是不同雨型的降雨诱发滑坡灾害的机制具有明显的差异性,一直以来确定累积降雨量时数是滑坡灾害气象预警的一个难点,逻辑回归模型可以通过显著性检验,自动去除相关性差的因子从而解决了这一问题。

4.1 滑坡灾害与降雨量统计分析

本文通过对2008-2013年有准确降水量的7872个历史滑坡灾害点当日和前期降水量进行统计,初步确定滑坡灾害发生当日、前期降水量的规律。如图10所示,有5160个（约占总滑坡灾害的65%）历史灾害当日降水量对累积降水量（7日累计降水量,下同）的贡献超过40%,其中1395个（约占总滑坡灾害的18%）历史灾害当日降水量对累积降水量的贡献率超过80%,说明四川省滑坡灾害受当日降水量影响很大,历史灾害的当日降水量在100~200 mm频次最多;有2093次滑坡灾害（约占总滑坡灾害的27%）前一日降水量对累积降水量的贡献率超过40%,滑坡灾害次数随降水量贡献率增大而减少,近50%的滑坡灾害前一日降水量超过50 mm,说明四川省滑坡灾害同样受前一日降水量影响;均有7%左右滑坡灾害前两日、前三日、前四日降水量对累积降水量的贡献率超过40%,前两日、前三日、前四日降水量集中分布在25 mm之下,说明四川省滑坡灾害与前两日、前三日、前四日降水有一定的关系,但是相关性较小;由此为了节约计算成本,本研究只提取滑坡灾害的当日降水和前四日降水建立滑坡灾害预警模型。
Fig. 10 Contribution of precipitation to cumulative precipitation on the day of the landslide hazard and the first four days

图10 滑坡灾害发生当日及前四日降水对累积降水量的贡献

4.2 滑坡灾害气象预警模型

基于四川省近4000个区域自动站小时降水数据,提取距历史灾害点最近的雨量站日降水量平均值,保证了滑坡灾害发生当日、前一日、前两日、前三日、前四日降水量的准确性。经处理后得到有准确降水量的历史灾害点7130个,其中4991个灾害点（70%)为建模数据,2139个灾害点（30%）为检验数据。以滑坡灾害危险性作为静态因子,概率化的降水量作为动态因子,滑坡灾害是否发生作为自变量建立逻辑回归模型（表3）,前三日、前四日概率化降水量未通过显著性检验,应剔除。
Tab. 3 Variables in the equation

表3 方程中的变量

B S.E. Wald df Sig. Exp(B)
滑坡灾害
危险性		 3.589 0.130 791.509 1 0.000 33.048
归day 11.545 0.368 1053.980 1 0.000 133 175.572
归day1 9.748 0.648 250.309 1 0.000 31 081.809
归day2 14.302 1.190 159.988 1 0.000 3 668 158.024
Constant -4.31 0.109 1739.165 1 0.000 0.011
剔除前三日、前四日概率化降水量后重新建模,根据SPSS分析得到的结果,Cox & Snell R Square为0.54,Nagelkerke R Square为0.72,表明模型总体拟合度很高;各因子均通过显著水平为0.05的Wald检验,且因子间相关系数很小,相互独立;模型对滑坡灾害发生的准确预测率达82.3%。建模结果如下：
P = EXP 3.589 H + 11.545 D 0 + 9.748 D 1 + 14.302 X 2 - 4.31 1 + EXP 3.589 H + 11.545 D 0 + 9.748 D 1 + 14.302 X 2 - 4.31 （4）
式中：D0为滑坡灾害发生当日降水概率化值;D1为滑坡灾害发生前一日降水概率化值;D2为滑坡灾害发生前两日降水概率化值;P为滑坡灾害发生概率,P越接近于1说明滑坡灾害发生的确定性越高。各因子Wald值如表4所示,结果表明当日降水量对滑坡灾害预警值影响最大,其次是滑坡灾害危险性,再次是滑坡灾害发生前一日降水量,滑坡灾害发生前两日降水对预警结果影响最小。该结果表明四川省滑坡灾害受集中性强降水影响较大,这与其降水特征吻合,从降水的角度验证了四川省滑坡灾害高发的原因。
Tab. 4 Wald value of different factors

表4 因子Wald值

因子 Wald
滑坡灾害危险性 791.509
归day 1053.980
归day1 250.309
归day2 159.988
Constant 1739.165

4.3 预警结果验证

通过对预警模型结果进行验证（表5）,有80%的滑坡灾害能成功预警,其中30%的滑坡灾害预警结果大于0.75,有18%的滑坡灾害预警结果大于0.99;有90%的特大型、大型滑坡灾害能成功预警,其中40%的特大型、大型滑坡灾害预警结果大于0.75,12%的特大型、大型滑坡灾害预警结果超过0.99。与现阶段四川省运行的概率化预警模型41%的预警准确率相比,漏报率大大减小。
Tab. 5 Statistics of the model verification

表5 模型验证统计表

P 滑坡灾害预警比例(%) 大型、特大型滑坡灾害预警比例(%)
≥0.5 80 90
≥0.75 30 40
≥0.99 18 12
滑坡灾害预警应达到空报率与漏报率的平衡,本研究针对未参与建模的群发性滑坡灾害个例检验模型的空报率和漏报率。据统计,2013年7月10日有群发性滑坡灾害在川东北地区发生,原有概率化模型预警结果与逻辑回归模型预警结果如图11、12所示。
Fig. 11 Probabilistic early warning model

图11 概率化预警模型

Fig. 12 Logistic regression model

图12 逻辑回归模型

概率化预警模型三级预警区几乎覆盖四川省中部、西北部所有区域,但这些区域并没有滑坡灾害发生,空报率非常大;二级预警区（即图中橙色区域）在西南区域出现误差,给实际预警工作的开展造成极大的不便;没有一级预警区,对当日大型、特大型滑坡灾害强度的预警准确性很低。
逻辑回归模型三级预警区相对集中;三级预警区面积较小,能包含当日所有滑坡灾害点,同时限制了空报率;二级预警区主要集中在滑坡灾害群发区,包含所有的大型、特大型滑坡灾害,从滑坡灾害强度预测上达到了提高;一级预警区面积很小但集中分布于大型、特大型滑坡灾害附近,对实际预警工作的开展起到了很好地指导作用。

5 结论与展望

本文在国内外滑坡灾害预警模型的基础上,结合四川省地质岩性特征、降雨特征以及两者的相关性分析,建立基于逻辑回归模型的滑坡长期监测预警模型—滑坡危险性区划,构建面向四川全省的显示滑坡灾害临灾气象预警模型,并对预警模型的结果进行验证,主要得到以下结论：
（1）四川省滑坡灾害具有强烈的空间分布特征,具体表现为沿龙门山断裂带、鲜水河断裂带、安宁河断裂带呈带状分布,在川东北地区呈集群性分布。
（2）当日降雨量对诱发滑坡灾害的贡献量最大,前一日次之,随着日期前移,降雨量对滑坡灾害发生的影响减弱。
（3）使用确定性系数的方法量化致灾因子,建立了致灾因子和滑坡发生与否的逻辑回归模型,完成了四川省滑坡灾害危险性区划。结果显示四川省滑坡灾害高危险性区域呈“Y”字型分布,此外川中、川东北滑坡危险性也非常高,这种分布与实际滑坡灾害发生情况相符。
（4）基于滑坡灾害危险性、当日降雨量、前一日降雨量、前两日降雨量因子和逻辑回归方法,建立了滑坡灾害临灾气象预警模型,通过2139个灾害点的验证,模型的预报准确率达到82%。并选取2013年7月10日作为个例,进行了滑坡灾害空间预警效果评价,结果表明本研究预警模型在提高预警精度的基础上缩小了预警范围,减少了空报率,整体预警精度上有了较大的提高。

The authors have declared that no competing interests exist.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
刘传正,刘艳辉.地质灾害区域预警原理与显式预警系统设计研究[J].水文地质工程地质,2007,34(6):109-115.通过总结分析国内外经验和新认识,文章把地质灾害区域预警原理初 步划分为隐式统计预警、显式统计预警和动力预警三种类型.隐式统计预警是把地质环境因素的作用隐含在降雨参数中,仅仅考虑降雨参数建立模型的方法.显式统 计预警是一种把地质环境因素变化与激发因素相迭加耦合而后建立判据模型的预警方法.动力预警是一种考虑地质体在降雨过程中自身动力变化过程而建立数学物理 判据方程的方法,实质上是一种解析预警方法.本文以系统工程思想为指导,比较全面地架构了显式统计预警系统的设计方法,包括需求分析、问题表述、目标设 定、设计原则、图层设计、系统建模、系统运算、预警等级确定、产品生成、产品会商、产品发布、预警校验、预警决策、应急行动、成本收益分析和系统完善升级 等内容.这项设计研究为具体研发设计某个地区的预警系统提供了思想平台,为建立国家、省和市(县)分层级联动的地质灾害区域预警体系提供了技术路线.

DOI

[ Liu C Z, Liu Y H.Early warning theory for regional geo-hazards and design of explicit statistical system[J].Hydrogeology & Engineering Geology, 2007,34(6):109-115. ]

[2]
刘艳辉,方志伟,温铭生,等.川东北地区强降雨诱发崩滑流灾害分析[J].水文地质工程地质,2014,41(2):111-115.川东北地区一直以来都是地质灾害高发地区,地质环境条件脆弱,极 端降雨事件频发,是区域地质灾害预警高度关注地区.以2012年6月下旬到7月上旬持续强降雨诱发崩滑流灾害过程为例,通过统计方法,从空间和时间角度, 分析地质灾害的空间分布、时间分布与降雨的相关关系,得到该区域降雨诱发地质灾害的规律特点及临界降雨判据.(1)川东北地区的降雨模式常见为数日持续降 雨,且降雨中出现数次(2～3日)的极值雨量,在这种降雨模式下,考虑到前期降雨的累积效应,需要着重关注第二次的极值降雨,往往会诱发更大的群发灾害过 程,并具有明显的滞后效应,此特点在巴中、达州、广元等地表现得尤为明显,而南充市的即时发灾特点更为明显.(2)川东北地区各市临界雨量不同,在少量前 期雨量累计情况下,广安最低,约为25 mm;其次是广元,约为50mm;最后是巴中、达州和南充,约为90mm.相关结论可为该地区的区域地质灾害预警研究和实践提供技术支撑.

[ Liu Y H, Fang Z W, Wen M S, et al.Analysis of geo-hazards and rainfall in northeast of Sichuan Province[J]. Hydrogeology & Engineering Geology, 2014,41(2):111-115. ]

[3]
侯圣山,李昂,周平根.四川雅安市雨城区地质灾害预警系统研究[J].地学前缘,2007,14(6):160-165.区域地质灾害的敏感性评价与诱发因素评价是区域群发性地质灾害预警预报的基础。文中以四川雅安市雨城区为例,系统介绍了区域地质灾害预警预报的方法。在区域地质灾害敏感性评价及区域降雨监测的基础上,研究了降雨诱发区域群发性地质灾害的规律,得出了地质灾害降雨阈值。研究了降雨诱发地质灾害预警预报的方法,建立了大中比例尺的地质灾害预警预报系统,并在四川雅安雨城区开展了系统运行。文中论述的方法可以用于县(市)级的地质灾害预警预报工作。模拟运行及实际运行效果表明,本方法效果较好,能够在类似区域进行推广。

DOI

[ Hou S S, Li Ang, Zhou P G.Preliminary study of the geohazard warning system based on weather forecasting and precipitation monitoring of Yucheng District,Ya an City, Sichuan Province[J]. Earth Science Frontiers, 2007,14(6):160-165. ]

[4]
李媛. 区域降雨型滑坡预报预警方法研究[D].北京:中国地质大学(北京),2005.

[ Li Y.Method for the warning of precipitation-induced landslides[D]. Beijing: China University of Geosciences,Beijing, 2005. ]

[5]
齐信,唐川,铁永波,等.基于GIS 技术的汶川地震诱发地质灾害危险性评价——以四川省北川县为例[J].成都理工大学学报:自然科学版,2010,37(2):160-165.在汶川地震诱发地质灾害应急调查的基础上,开展地震诱发地质灾害危险性评价.以北川县为例,应用GIS技术对5个主要评价指标进行提取和叠加分析,并结合降雨因子最终完成北川县灾后危险性评价图.评价结果表明:高危险区面积占52.6%,地质灾害点占95.9%;中危险区面积占25.1%,地质灾害点占3.7%;低危险区面积占22.3%,地质灾害点占0.4%.

DOI

[ Qi X, Tang C, Tie Y B, et al.Hazard assessment of geohazards triggered by the Wenchuan earthquake using GIS technology: Taking Beichuan county of Sichuan Province for example[J]. Journal of Chengdu University of Technology:Science & Technology Edition, 2010,37(2):160-165. ]

[6]
Cross M.Landslide susceptibility mapping using the matrix assessment approach: A Derbyshire case study[J]. Geological Society, London, Engineering Geology Special Publications, 1998,15(1):247-261.Abstract Civil engineering schemes such as new highways and railway lines, regional planning and large-scale land-management projects in areas known to have a landslide problem require regional landslide susceptiblity evaluation. The Matrix Assessment Approach (MAP) is introduced as a medium-scale landslide hazard mapping technique for establishing an index of slope stability over large areas. The method allows the relative landslide susceptibility to be computed over large areas using a discrete combination of geological/geomorphological parameters. MAP was applied to a region in the Peak District, Derbyshire. The model identified key geological/geomorphologial parameters involved in deep-seated failures, provided an effective means of classifying the stability of slopes over a large area and successfully indicated sites of previously unmapped landslides. The resultant regional landslide susceptibility index provides useful preliminary information for use at the desk study and reconnaissance stages of large-scale civil engineering works such as highway construction.

DOI

[7]
唐川,朱静.GIS 支持下的地震诱发滑坡危险区预测研究[J].地震研究,2001,24(1):73-81.

[ Tang C, Zhu J.GIS based earthquake triggered landslide hazard prediction[J]. Journal of Seismological Research, 2001,24(1):73-81. ]

[8]
陈晓利,祁生文,叶洪.基于GIS 的地震滑坡危险性的模糊综合评价研究[J].北京大学学报:自然科学版,2008,44(3):434-438.

[ Chen X L, Qi S W, Ye H.Fuzzy comprehensive study on seismic landslide hazard based on GIS[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2008,44(3):434-438. ]

[9]
Conoscenti C, Di Maggio C, Rotigliano E.GIS analysis to assess landslide susceptibility in a fluvial basin of NW Sicily (Italy)[J]. Geomorphology, 2008,94(3):325-339.Landslide hazard assessment, effected by means of geostatistical methods, is based on the analysis of the relationships between landslides and the spatial distributions of some instability factors. Frequently such analyses are based on landslide inventories in which each record represents the entire unstable area and is managed as a single instability landform. In this research, landslide susceptibility is evaluated through the study of a variety of instability landforms: landslides, scarps and areas uphill from crown . The instability factors selected were: bedrock lithology, steepness, topographic wetness index and stream power index. The instability landform densities computed for all the factors, which were arranged in Unique Condition Unit, allowed us to derive a total of three prediction images for each landslide typology. The role of the instability factors and the effects generated by the use of different landforms were analyzed by means of: a) bivariate analysis of the relationships between factors and landslide density; b) predictive power validations of the prediction images, based on a random partition strategy. The test area was the Iato River Basin (North-Western Sicily), whose slopes are moderately involved in flow and rotational slide landslides (219 and 28, respectively). The area is mainly made up of the following complexes: Numidian Flysch clays (19%, 1%), Terravecchia sandy clays (5%, 1%), Terravecchia clayey sands (3%, 0.3%) and San Cipirello marly clays (9%, 0%). The steepness parameter shows the highest landslide density in the [11–19°] class for both the typologies (8%, 1%), even if the density distributions for rotational slides are right-asymmetric and right-shifted. We obtained significant differences in shape when we used different instability landforms. Unlike scarps and areas uphill from crowns, landslide areas produce left-asymmetric and left-shifted density distributions for both the typologies. As far as the topographic wetness index is concerned, much more pronounced differences were detected among the instability landforms of rotational slides. In contrast, the flow landslides produce normal-like density distributions. The latter and the rotational slide landslide areas produce the highest density values in the class [5.5–6.7], despite an abrupt decreasing trend starting from the first class [3.2–4.4], which is generated by the density values of the rotational slide scarps and areas uphill from crowns. The stream power index at the foot of the slopes, which was automatically derived using a GIS-procedure, shows a positive correlation with the landslide densities marked by the maximum classes: [4.8–6.0] for flows, and [6.0–7.2] for rotational slides. The validation procedure results confirmed that the choice of instability landform influences the results of the susceptibility analysis. Furthermore, the validation procedure indicates that: a) the predictive models are generally satisfactory; b) scarps and zones uphill from crown areas are the most diagnostically unstable landforms, for flow and rotational slide landslides respectively.

DOI

[10]
唐川,朱大奎.基于GIS 技术的泥石流风险评价研究[J].地理科学,2002,22(3):300-304.

[ Tang C, Zhu D K.Assessment of debris flow risk of Yunnan province by using GIS[J]. Scientia Geographica Sinica, 2002,22(3):300-304. ]

[11]
兰恒星, 伍法权. 基于GIS 的云南小江流域滑坡因子敏感性分析[J]. 岩石力学与工程学报,2002,21(10):1500-1506.进行滑坡因子敏感性分析是滑坡研究的一个重要的步骤,有利于研究滑坡的变形破坏机理及分析滑坡的空间分布规律。采用滑坡确定性系数的方法对云南小江流域进行了影响因子敏感性分析,确定了有利于滑坡发育的岩性、结构、坡度、高程、断裂等影响因子。

DOI

[ Lan H X,Wu F Q.Analysis on susceptibility of GIS based landslide triggering factors in Yunnan Xiaojiang watershed[J]. Chinese Journal of Rock Mechanics and Engineering, 2002,21(10):1500-1506. ]

[12]
兰恒星,伍法权. GIS支持下的降雨型滑坡危险性空间分析预测[J].科学通报,2003,48(5):507-512.基于GIS的滑坡分析模型的基础上, 针对现有模型的优势和不足, 提出了改进的SINMAP模型. 该模型将滑坡确定性模型与基于DEM的水文分布模型进行有效的集成, 考虑了由于降雨引起的地下水的不同分布, 所造成的静水压力及地下水渗流过程中产生的动水压力对滑坡稳定性的影响. 以中国西南典型的降雨型滑坡分布区--云南小江流域为例, 探讨了模型应用的有效性, 详细研究分析了小江流域滑坡危险性在不同降雨条件下的空间分布规律, 特别是随降雨条件的变化、滑坡变形失稳区域的扩展趋势以及失稳位置、失稳面积等的空间变化特征, 给出了滑坡发生与降雨、地形坡度、集水区面积等因素之间的定量关系. 在此基础上, 推算出小江流域滑坡失稳的降雨量阈值, 对区域性降雨型滑坡危险性进行快速有效的分析和预测.

DOI

[ Lan H X,Wu F Q.Spatial evaluation and forecast landslide risk based on GIS[J]. Chinese Science Bulletin, 2003,48(5):507-512. ]

[13]
王济川,郭志刚.Logistic 回归模型:方法与应用[M].北京:高等教育出版社,2001.

[ Wang J C, Guo Z G.Logistic regression models: methods and application[M]. Beijing: Higher Education Press, 2001. ]

[14]
许冲,徐锡伟.逻辑回归模型在玉树地震滑坡危险性评价中的应用与检验[J].工程地质学报,2012,20(3):326-333.At 07: 49(Beijing time)on April 14, 2010,a catastrophic earthquake with Ms 7.1 struck Yushu County,Qinghai Province,China.2036 landslides were interpreted from aerial photographs and remote sensing images,verified by selected field checking.The spatial distribution of the landslides was evidently strongly controlled by the locations of the main surface fault ruptures.The landslides commonly occurred close together.Most of the landslides were small.The landslides were of various types.They were mainly shallow,disrupted landslides,but also included rock falls,deep-seated landslides,liquefied landslides, and compound landslides.The aim of this study was to apply and validate logistic regression model for the Yushu earthquake triggered landslide hazard mapping.The Geographic Information Systems(GIS) and Remote Sensing(RS)technologies were used.A spatial database was developed and analyzed using GIS technology.It includes the landslides and associated controlling parameters which may have influence on the occurrence of landslides..The twelve factors that influence landslide occurrence were created in raster data format base on GIS platform.They include elevation,slope angle,slope aspect,slope curvature,slope position,drainages,lithology,faults,roads,normalized difference vegetation index(NDVI),co-seismic main surface fault-ruptures, and peak ground acceleration(PGA).the coefficients of the evaluation variables were estimated using logistic regression model and were used to calculate the landslide hazard index for the entire study area within a GIS environment.The validation result showed a success rate of 83.21% between the hazard map and the locations of Yushu earthquake triggered landslides.The landslide hazard assessment rank result showed that the"very high hazard"level covers about 4.97% of the total study area but has a number of 766 landslides(37.62% of the total landslides number),which representing good quality of the landslide hazard map.The sequence of the landslide number density percentage descends expeditiously accompany with the hazard level decreasing.

DOI

[ Xu C,Xu X W.Logistic regression model and ITS validation for hazard mapping of landslides triggered by Yushu earthquake[J]. Journal of Engineering Geology,2012,20(3):326-333. ]

[15]
许冲,戴福初,徐素宁,等.基于逻辑回归模型的汶川地震滑坡危险性评价与检验[J].水文地质工程地质,2013,40(3):98-104.以2008年5月12日汶川地震区为研究区,基于高分辨率航片与 卫星影像开展地震滑坡目视解译,制作了汶川地震滑坡编录图.选择坡度、坡向、高程、与水系距离、与公路距离、与映秀-北川断裂距离、地震烈度、岩性共8个 影响因子开展地震滑坡危险性评价工作.滑坡样本采用前期48007处滑坡编录点数据,不滑样本为在基于证据权重模型的滑坡危险性评价结果的低危险区与极低 危险区随机选择的48000个点.基于这8个影响因子与逻辑回归模型,建立了汶川地震滑坡危险性索引图.采用这48007个滑坡样本点与汶川地震滑坡最新 编录的增加滑坡,分别进行模型的成功率与预测率检验.结果表明,模型成功率为81.739％,预测率达到86.278％.

[ Xu C,Dai F C,Xu S N,et al.Application of logistic regression model on the Wenchuan earthquake triggered landslide hazard mapping and its validation[J]. Hydrogeology & Engineering Geology, 2013,40(3):98-104. ]

[16]
刘明学,陈祥,杨珊妮.基于逻辑回归模型和确定性系数的崩滑流危险性区划[J].工程地质学报,2014,22(6):1250-1256.This paper aims at finding out the functional relationship between landslide and impact factors, using statistics analysis under logistic regression model and certainty factor. It tries to draw the block map of the hazard zonation according to the level of risk of landslide occurrence in Guizhou Province using GIS technology. First, a calculation of the certainty factor of landslide occurrence is done according to the area of occurred disasters in impact factor subset and the area of impact factor subset. Secondly, the possibility of landslide occurrence is defined as dependent variable. The certainty factor of occurring disaster in subset is defined as independent variable. Then, an analysis is done to find out the functional relationship between them under logistic regression model. Thirdly, a calculation of the probability of landslide occurrence is initiated. It tries to divide the research area into 10 risk level sub-areas according to the result with the aim to draw block map of the hazard zonation of landslide. Finally, an evaluation on the hazard zonation of landslide is carried out. The result shows that the hazard zonation method, based on logistic regression model and certainty factor for landslide, is effective.

DOI

[ Liu M X, Chen X, Yang S N.Zonation of landslide risk with logistic regression model and certainty factor[J]. Journal of Engineering Geology, 2014,22(6):1250-1256. ]

[17]
Heckerman D.Probabilistic interpretations for MYCIN's certainty factors[C]. Conference on Uncertainty in Artificial Inteligence, AUAI press, 1985:9-20.

[18]
Shortliffe E H, Buchanan B G.A model of inexact reasoning in medicine[J]. Mathematical biosciences, 1975,23(3):351-379.Medical science often suffers from having so few data and so much imperfect knowledge that a rigorous probabilistic analysis, the ideal standard by which to judge the rationality of a physician's decision, is seldom possible. Physicians nevertheless seem to have developed an ill-defined mechanism for reaching decisions despite a lack of formal knowledge regarding the interrelationships of all the variables that they are considering. This report proposes a quantification scheme which attempts to model the inexact reasoning processes of medical experts. The numerical conventions provide what is essentially an approximation to conditional probability, but offer advantages over Bayesian analysis when they are utilized in a rule-based computer diagnostic system. One such system, a clinical consultation program named mycin, is described in the context of the proposed model of inexact reasoning.

DOI

[19]
张锡涛,刘翔宇,谢谟文,等.基于岩质滑坡引发泥石流的影响范围评价模型[J].工程地质学报,2013,21(4):598-606.Debris or mud flow caused by the landside is one of complicated geologic hazards in mountain area, which is usually related to geology, lithology, the mechanical characteristics of the rock and soil, rainfall, groundwater, and land usage condition. The analysis of immanent relationships between landslide and mud flow can not only afford a reference for the analysis of landslide failure mechanism, but also be used as the basis of evaluation of mud flow triggered by landslides. In this paper, the depth integral is applied two two-dimensional mathematical model of mud flow. The model is based on the principle of conservation of mass and viscous Newtonian fluid Navier-Stovkes equation. Then, this equation is numerically computed by using the finite difference method. The relational expression between streams tilt angle and the width of affected range of mud flow is obtained from statistics. It is applied to analyzing the possibility of mud flow induced by landslides which once failed under the similar geological conditions. Combining with GIS,the model can also be used to predict impacted range of mud flow by using risk map to show the zones that may be affected by the mud flow.

DOI

[ Zhang X T, Liu X Y, Xie M W, et al.Mathematical model for evaluating affected range of debris flow induced by rock landslide[J]. Journal of Engineering Geology,2013,21(4):598-606. ]

[20]
常鸣,唐川,苏永超,等.雅鲁藏布江米林段泥石流堆积扇危险范围预测模型[J].工程地质学报,2012,20(6):971-978.The Milin county along Yarlungzangbo river is closed to the Indian plate with the Eurasian plate collision zones, where the tectonic movement is intense and the seismic activity is frequent. So debris flow are widely distributed. This paper uses the remote sensing and field investigation and examines 34 debris flow gullies which have an integrated deposit fan. Using geographical information system and remote sensing software, the paper analyses the basic parameters of the basin such as basin relief and the volume source. The regression analysis on the maximum length and width runout and their topographical factors is conducted using the software Mat lab. It is found that the maximum length and width runout have an obvious index correlation with the basin relief and the volume source. Based on the found single factor regular, the predicting model on evaluating the debris flow's hazardous range is established. This model absorbs the previous model advantages, distinguishes the risk range by less factors and saves time. It can be better and faster against the disaster.

DOI

[ Chang M, Tang C, Su Y C, et al.Prediction model for debris flow hazard zone on alluvial fan in Milin section of Yarlungzangbo river, Tibet[J]. Journal of Engineering Geology, 2012,20(6):971-978. ]

Options
Outlines

/

〈 〉