IMHE OpenIR  > 山地灾害与地表过程重点实验室
面向泥石流短临预报的降水监测方法
Alternative TitleRainfall observation method for short-term debris flow forecast
史朝
Subtype博士
Thesis Advisor韦方强
2018
Degree Grantor中国科学院大学
Place of Conferral北京
Degree Discipline自然地理学
Keyword泥石流 规模 频率 危险性 地震
Abstract泥石流短临预警与预报是减轻地质灾害损失的一项重要非工程措施。本研究讨论了面向泥石流短临预报的降水监测方法所面临的关键挑战,对比分析了目前复杂地形下多普勒天气雷达与雨量计两种常用降水监测方法的优势与不足,设计了一系列在复杂地形条件下多普勒天气雷达数据质量控制算法,建立基于电波传播特性与数字高程数据(DEM)数据的地物回波模拟、遮挡订正以及混合扫描等规避地物影响的算法;形成了以“显著垂直反射率因子廓线”(AVPR)为基础的订正方法用于修正降雨落区的反射率因子;设计了基于距离和高度加权的雷达拼图以扩大雷达的观测范围;基于层性降水与对流性降水的识别进行了“雨强—反射率因子”(R-Z)关系估测及降雨量累积;为降低雷达估测场与真实降雨场的偏差,结合卡尔曼滤波器对雷达估测的雨量进行了偏差订正;将降水估测雨区分成了两种类型,分别评估了雷达估测降水量的精度;基于降水型泥石流的“雨强—持续时间”(I-D)经验模型,分析对比了分别由雷达与雨量计观测数据导出的I-D模型;在以上此基础上,进一步讨论分析了降水空间差异及观测误差对确定泥石流发生概率的影响;结合“连续性交叉相关跟踪”(COTREC)算法设计了复杂地形下雷达定量降水预报算法,并进行了效果验证;通过对泥石流灾害成因的分析,基于可拓学方法建立了中小尺度区域的泥石流短临预报模型,阐述了激发泥石流的静态和动态影响因子中各类数据的获取方法,建立了相应的物元模型,并结合多普勒天气雷达实时的雨量估测及外推的雨量预报,实现了面向潜势泥石流流域的可拓预报模型。本文选择了受汶川地震严重影响的强扰地带作为本文的研究区域,包含汶川县、理县、茂县、平武县、北川县、安县、绵竹市、都江堰市、彭州市、什邡市、宝兴县、芦山县以及绵阳市与江油市部分地区。震后每年汛期雨季,特别是6至8月份,该地区泥石流频发,地质灾害损失惨重。该区域内有成都与绵阳两部新一代多普勒天气雷达能够实现对该区域的观测覆盖,考虑到该地区内的雨量计布网密度差异较大,在开展雷达定向估测及预报降水的基础上,进一步分析降水空间差异与观测误差对确定泥石流发生概率的影响。并在此基础上,利用统计方法建立了该地区的泥石流短临预报的标准物元模型,确定了预报模型中的各类参数以及各参数的域值、节域和权重系数。基于可拓学模型并“地理信息系统”(GIS)技术研究设计了泥石流短临预报算法,并对2012~2014年6次降水过程进行了泥石流预报。通过以上研究,取得了如下研究成果:(1)建立了适应川西地区的雷达定量估测降水的质控算法与属地化参数,包括川西地区地物回波模拟,降水经验关系,以及雷达估测降水的外推预报。(2)分析了降水空间变化与观测误差对于确定泥石流发现概率的影响,评估了累积雨量(E)、持续时间(D)、平均雨强(I)三个与降水特征相关的因子与观测取样距离(降水观测取样点到泥石流发现位置之间的距离)的关系,三个因子随取样距离的相对变化关系呈现单调递增且指数小于1的幂律关系,10km范围内由于降水空间变化引起E、D、I三个因子的不确定性分别小于45%、43%、58%。这为泥石流成因预报建立雨量相关物元,确定阈值及权重提供了依据。(3)以该区域泥石流形成环境背景和泥石流灾害数据库为基础,利用可拓学方法确定了泥石流短临预报模型中的各参数,建立了基于潜势泥石流流域的泥石流短临预报模型。(4)将GIS技术与区域泥石流短临预报相结合,基于多普勒天气雷达估测与预报,建立了以雷达遥感技术为核心的泥石流短临预报应用模型,滚动预报保持与雷达时间分辨率一致,5~6分钟,预报时长为1h。对2012~2014年6次降水过程进行了泥石流预报的结果表明,该模型命中率最高为93%,最低空报率为31%。本研究创新点主要表现在以下几个方面:(1)开展了面向川西泥石流多发地区的雷达定量估测降水及质控算法研究,为泥石流短临预报提供高精度和高分辨率的定量估测及预报降水产品。(2)结合I-D模型,使用雷达及雨量计观测数据,研究分析了降水空间变化及观测误差对确定累积雨量、降水持续时间、平均雨强的潜在不确定性,并进一步使用雷达估测雨量结果建立了研究区域内的I-D经验关系。(3)将泥石流形成因素与可拓学模型相结合,研究建立了面向潜势泥石流流域的泥石流短临预报模型。并基于雷达的高时空分辨率优势,实现了泥石流短临预报的高频次滚动预报,滚动预报间隔以雷达定量估测及预报产品的更新时间(5~6分钟)为最小时间单元。本研究为今后进一步利用多参数天气雷达开展降雨型泥石流预报提供了一些依据和借鉴,有助于促进在减灾领域中更多地利用先进的大气探测技术和天气预报产品。
Other AbstractShort-term warning and forecast for debris flow is an important non-engineering measure to reduce the loss of geological disasters. This study discusses the key challenges faced by precipitation monitoring methods for prediction of debris flows. The advantages and disadvantages of the two commonly used precipitation monitoring methods, Doppler weather radar and rain gauge, are compared. A series of improving measures for Doppler weather radar data quality control are considered. In relation to the complexity of terrains, an algorithm based on the characteristics of ground and electromagnetic wave propagation are designed to simulate the clutter echoes and calculate the blockage ratio with the DEM. A hybrid scan mode for avoiding ground effects is established; A reflectivity profile based on AVPR is formed. The revised method is used to adjust the reflectivity factor of rainfall falling zone; The mosaic for multiple radars is applied with distance and height-weighted value; The R(Z)-based rainfall rate estimation and accumulation are performed based on the recognition of stratiform and convective precipitation; To decrease the bias between the radar-based rainfall estimate and rain gauge observation, a Kalman filter is applied to correct the hourly rainfall estimation; An Establishment of the ID model based on radar estimation is done and and compared with models deduced by rain gauges obervation; On this basis, the paper further discusses the impact of precipitation spatial differences and observational errors on the probability of debris flow occurrence; Combining the COTREC algorithm, a radar quantitative precipitation forecasting algorithm is configured for the complex terrain and validation is performed; Based on the regional short-term prediction model for debris flows at medium and small scales, various types of static and dynamic factors forming debris flow have been described. A corresponding matter-element model has been established and combined with Doppler weather radar in real time. Rainfall estimation and extrapolated rainfall forecasting are input into the Extenics forecasting model for potential debris flow watersheds.This article selected the severely disturbed zone affected by the Wenchuan earthquake as the study area of this article, including Wenchuan County, Lixian County, Maoxian County, Pingwu County, Beichuan County, Anxian County, Mianzhu City, Dujiangyan City, Pengzhou City, Shifang City, and Baoxing County. County, Lushan County, Mianyang City and parts of Jiangyou City. After the earthquake, during the annual rainy season, especially in June-August, a high occurrence of debris flow over this region happened and caused vast losses in geological disasters. In addition, two new generations of Doppler weather radars in Chengdu and Mianyang can cover this area. Considering the large spatial density differences of rainfall gauge networks within this area, radar-based precipitation estimation and forecasting are carried out. Based on the above analysis, the impact of precipitation spatial difference and observation error on the probability of occurrence of debris flow is analyzed. A standard matter-element model for short-term prediction of debris flows in the region are established using statistical methods. And the various parameters in the forecast model, as well as the domain values, domain ranges, and weight coefficients of the parameters are determined. The short-term prediction algorithm for debris flows is studied and designed using GIS technology, and debris flow forecast was made for the six precipitation processes from 2012 to 2014. Through the above studies, the following research results have been obtained:(1) The quality control algorithm and localized parameters for quantitative precipitation estimation in the western Sichuan region were established, including the simulation of ground echoes in western Sichuan, the empirical relationship between rainfall and radar reflectivity were obtained.(2) The impact of precipitation spatial difference and observation error on the occurrence probability of debris flow were analyzed using I-D model, and the relative change for the accumulated rainfall, duration, rainfall ratio versus distance were discussed. For the statistical relationship for the three parameters versus distance, the monotonically increasing and exponentially less than one of power-law relationship appears with the relative change of the distance, which provides a basis for establishing rainfall-related matter elements for the prediction of debris flows and the determination of thresholds and weights.(3) Based on the background of the formation of debris flow in the area and the database of debris flow disasters, the parameters of the short-term prediction model for debris flow were determined by statistical methods, and an Extenics model for short-term prediction of debris flows based on the potential watershed was established.(4) Combining GIS technology with regional debris flow short-term prediction, radar-based rainfall estimation and forecasting, an application model for short-term prediction of debris flows based on radar detection technology was established, and The timing interval of rolling prediction nearly equals to the radar scan temporal resolution of 5-6 minutes, and the forecast period is one hour. The results of the debris flow forecast for the six precipitation processes from 2012 to 2014 shown that the highest hit rate of this model is 0.93, and the lowest false alarm rate is 0.31.The innovations of this research mainly include the following aspects:(1) The study on radar quantitative precipitation and quality control algorithms for debris flow-prone areas in western Sichuan was conducted to provide accurate and high-resolution quantitative precipitation estimate and forecasting products for short-term prediction of debris flows.(2) In combination with the I-D model and using radar and rain gauge observation data, the potential impact of precipitation spatial varations and observational errors on cumulative rainfall, duration of precipitation, and mean rainfall intensity were studied and analyzed, and furthermore, the radar-based rainfall estimation was used to estimate an I-D empirical relationship for the study area.(3) Applying exentic model with the main factors of debris flow formation, a short-term prediction model for debris flows for the potential watershed was established. Based on the advantage of radar's high spatial-temporal resolution, high-frequency rolling short-term prediction for debris flow was implemented. The rolling prediction interval is decreased to the update time (5-6 minutes) of the radar-based forecast productThis study built some basis for the further use of weather radar for forecasting the rainfall-type debris flow and would also help promote the more usage of advanced atmospheric detection technologies and weather forecast products in the field of disaster reduction. 
Pages134
Language中文
Document Type学位论文
Identifierhttp://ir.imde.ac.cn/handle/131551/24782
Collection山地灾害与地表过程重点实验室
Affiliation中国科学院成都山地灾害与环境研究所
First Author Affilication中国科学院水利部成都山地灾害与环境研究所
Recommended Citation
GB/T 7714
史朝. 面向泥石流短临预报的降水监测方法[D]. 北京. 中国科学院大学,2018.
Files in This Item:
File Name/Size DocType Version Access License
面向泥石流短临预报的降水监测方法.pdf(10912KB)学位论文 开放获取CC BY-NC-SAView Application Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[史朝]'s Articles
Baidu academic
Similar articles in Baidu academic
[史朝]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[史朝]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 面向泥石流短临预报的降水监测方法.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.