【作者】 刘志辉；

【导师】 高井祥；

【作者基本信息】 中国矿业大学 ， 地图制图学与地理信息工程， 2009， 博士

【摘要】 我国(尤其是新疆)是一个积雪广布、雪灾和春洪多发的国家(地区)。在全球气候变化的大背景下,新疆洪水灾害尤其是冰、雪融水洪水频次存在增大的趋势,而且洪灾损失巨大。可以讲,任何灾害的预防都应该建立在预警基础上。只有弄清融雪机制、水量及过程,准确的监测、预测和预警,制定科学有效的防范措施,才能避免灾害或将灾害损失降低到最低限度。而且,遥感(RS)、地理信息系统(GIS)以及全球定位系统(GPS)即“3S”技术和计算机技术以及大气数值预报模式,为积雪信息的大面积动态监测、分布式融雪径流模型和融雪洪水预警预报决策支持系统的建立提供了有力的基础和保障。因此,对积雪和融雪信息实时提取、模型分析和计算,及时、准确地向政府部门提供融雪洪水预警预报,对于防灾减灾具有重要意义,同时也可以产生巨大的社会、经济和生态效益。本研究结合新疆春季融雪及洪水过程的特点,基于“3S”技术以及较高分辨率DEM的流域下垫面和积雪信息提取研究,利用大气模式WRF V2.2,采用国家气象局T213L13预报场数据,对典型研究区(天山北坡军塘湖流域)2008年融雪期进行24h短期气象场预报,以及借助于GPS进行同步野外现场的气象、水文和积雪观测获得了第一手数据,为融雪径流模型的分布式输入奠定了基础。利用WRF预报出的气象场和积雪、融雪数据驱动分布式新疆融雪径流模型,对融雪洪水过程进行短期预报。并利用新疆融雪洪水预警模式计算评价洪水的灾害级别和预警级别。通过新疆融雪洪水预警决策支持系统(DSS)为决策者提供融雪洪水预警,辅助决策者生成决策方案,为正确的防洪减灾决策提供有利的技术和信息支持。本研究基于“3S”技术着重对新疆融雪洪水预测预警以及决策支持进行研究,提出了相应的思路和方法,研究的内容和结论如下:(1)基于GIS和计算机技术,在融雪洪水预警DSS中开发研制了GIS数据的空间分析模块包括流域信息提取子模块,使得分布式流域信息提取,如DEM、坡度、坡向、流向、水系等数字地图,不必依赖于其它GIS软件即可完成,而且,时间和空间分辨率可根据需要灵活调整,在新疆典型研究区的应用中精度和效果比较理想。(2)基于易于获取的高时间分辨率的MODIS遥感影像和GIS技术对研究区积雪和融雪信息进行了定量提取,并借助于GPS进行了同步野外现场观测,予以分析和检验。结果表明,经“雪盖系数(SF)”提取的雪盖信息较经“归一化雪盖指数(NDSI)”提取的雪盖信息精度要高,SF提取雪盖面积的平均精度为相对误差8%以内。在积雪雪深进行分级的基础上建立了雪深与MODIS的CH1和CH3的双线性回归方程,即新疆稳定积雪期积雪深度的反演模型,经检验,平均绝对误差为1.47cm,相对误差的平均值为10.96%。但在融雪期中该模型不适宜使用。(3)应用覃志豪等人的劈窗算法,由MODIS数据的解译和计算得到了典型研究区融雪期内的地表温度(LST)结果,经过检验,误差在0.5～3.0℃之间,经过局部修正基本可以满足模型输入的需求。(4)利用大气模式WRF V2.2,采用国家气象局T213L31预报场作为初始场及侧边界条件,选择适合的物理方案,实现了研究区2008年融雪期的24h数值天气预报。预报的空间分辨率为1km,时间步长为1h。预报的结果与实测基本相符,满足精度要求。(5)在“3S”技术支持下,自主设计和建立了基于能量平衡和水量平衡的分布式融雪径流模型。设计思路上以RS(ETM、MODIS)和现场观测数据为主要数据来源,考虑模型参数的时空分布差异,对其进行空间分布式处理。模型应用时,在新疆典型研究区WRF预报出的气象场数据(气温、相对湿度、风速、辐射)和现场积雪、融雪观测数据的基础上,驱动分布式融雪径流模型,做出融雪洪水过程的短期预报。应用和验证结果表明,预报的径流过程与实际观测值基本吻合,可为今后水文预报及无资料或少资料区水资源管理以及气候变化研究提供重要参考。(6)结合新疆当地实际建立了适用于新疆融雪洪水的预警模式和预警标准。在融雪洪水预警指标研究中,既考虑融雪洪水发生的大小量级和时空分布,还结合洪水可能发生的地区的经济社会状况,选择能够反映融雪洪水灾害的主要因子,即致灾因子、承灾因子和防洪设施贡献因子。并使用经过量化处理后的预警指数来综合反映区域融雪洪水的风险大小,以此作为划分春季融雪洪水预警等级的依据。可根据预测的洪峰和洪量,按照灾害的严重性和紧急程度,提前进行融雪洪灾预警。(7)针对新疆融雪洪水的特点和预警决策需求,提出了融雪洪水预警DSS的结构,并且在B/S和J2EE的结构下提出了该DSS的总体框架和软件实现方法,包括DSS的总体框架设计、基于B/S和J2EE结构的DSS的软件开发环境以及程序设计方法的选择等。应用实例表明,该DSS为模型参数的自动获取、模型之间的数据交换、模型结果的可视化表达等提供了有效的工具。实现了J2EE技术在融雪洪水预警DSS中的应用,设计效果良好。本文最后,作者对本文的研究进行了总结,并提出进一步研究的一些关键问题及其研究的发展方向。更多还原

【Abstract】 China,especially Xinjiang, is a country or region with wide snow distribution, snow disaster and spring flood frequently-occurring. Under the background of global climate change there is a tendency to increase the flood disaster frequency, particular glacier and snow melt floods, whose loss is very large. The prevention of each disaster should be based on an early-warning. Only to ravel the snowmelt mechanism, runoff and processes, to monitor, predict snowmelt flood and early-warning, to institute some effective measures before the disaster could be avoided or reduced in the lowest. Remote Sensing (RS), Geographical Information System (GIS), Global Positioning System (GPS), Compute technology and atmospheric numeric forecasting mode provide a powerful foundation and guarantee for monitoring the snow information in larger area and setting up the distributed snowmelt runoff model and the DSS for snowmelt flood early-warning. Therefore it has an important significance and a great deal of social, economic and ecological benefits for disaster prevention and reduction to extract the information of snow and melt, analyze and calculate with models, provide the flood early-warning to government in real time and precision.This research made the extraction of the information of underlying surface and snow in the study area (Juntanghu watershed in the northern slope of Tianshan Mountain) based on‘3S’technology and higher resolution DEM according to the features of the spring snowmelt flood in Xinjiang. This study using the atmospheric mode WRF2.2 and Chinese national bureau T213L13 data 24h meteorological field was forecasted from Feb. to March 2008 in the study area. Then the distributed Xinjiang snowmelt runoff model was drived by the forecasted meteorological field from WRF and snow melt data to forecast snowmelt flood processes. And the disaster and early-warning grade of floods were calculated and assessed by the early-warning model of snowmelt flood. With the DSS for snowmelt flood early-warning in Xinjiang the early-warning could be provided for decision maker to assist decision making, to provide the technical and informational support for disaster prevention and reduction.This study paid attention to the modeling snowmelt flood and decision making for early-warning and put forward correspondent ideas and approaches. The study contents and conclusions are following: (1) Based on‘3S’and computer technology this study developed a GIS spatial analysis module including sub-module of watershed information extraction in the DSS for snowmelt flood early-warning. It finished the task extracting watershed information, such as DEM, slope, aspect of slope, flow direction, river network etc., is not depend on the other GIS software, and the spatial and temporal resolution may be adjusted by need as far as possible. The results are good.(2) It was analyzed and verified that the snow and melt information extracted from the MODIS image data with higher temporal resolution and by GIS technology, meanwhile, by means of GPS to investigate it in local field. The results indicated: the precision of snow cover extracted by snow fractional (SF) is more than that of normalized difference snow index (NDSI). The average precision of snow cover by SF is below 8% of relative error. On the foundation of grading snow depth the double linear regression model which inversed the snow depth during stable snow period was set up. Through a test the average absolute error is 1.47cm, the mean relative error is below 10.96%. But this model is unsuited for snow melting period.(3) According to the Split Window Algorithms of Qin Zhihao and so on the land surface temperature (LST) during snow melting period in study area was obtained by the interpretation and calculation from MODIS data. By test the error is from 0.5 to 3.0℃, and if the algorithm is modified locally the results can be met the need of model input.(4) The limited-region 24-hour Numeric Weather Forecasting System is established by using the new generation atmospheric model——Weather Research and Forecasting Model (WRF) V2.2 with the initial fields and lateral boundaries provided by the T213 L31 from Chinese National Meteorological Bureau. The predicted spatial resolution is 1km, the temporal is 1h. The predicted results accorded with the observed on the whole, have got a good effect.(5) The self-design distributed snowmelt runoff model based on energy balance, water balance and‘3S’technology was established in this study. It is designed that taking RS(MODIS,ETM)data and the observed data on spot as the main data sources, model parameters with spatial and temporal difference should be processed in spatial distributed form. In application the distributed Xinjiang snowmelt runoff model was drived by the forecasted meteorological field from WRF and snow melt data to forecast snowmelt flood processes. The results indicate: the predicted runoff process fitted close to the observed.(6) This study established an early-earning mode and standard which is adequate for the local situation. In the study of early-warning index the main factors which can affect snowmelt flood disaster were selected, such as causing disaster factor, bearing disaster factor and the contributive factor of facilities for flood control, for not only the magnitude, grade and spatial and temporal distribution of snowmelt flood should be considered, but also the local economic and social situation should be integrated. The quantitative early-warning index which reflects comprehensively the risk size of local flood regards as the basis to grade the early-warning of spring snowmelt flood in Xinjiang.(7) To meet the need of early-earning decision this study designed the structure of Decision Support System (DSS) for snowmelt flood early warning in Xinjiang. Based on the structure of B/S and J2EE, the general frame and the methods of software realization in the DSS were put forward, including design of general frame, the selection of development environment and programming way. The applied example shows: this DSS is an effective tool for automatic obtaining of model parameters, data transition between models, visualization of model results. And the application of J2EE technology in the DSS was realized. The effect of design is good.At the end of the dissertation, the author draws a conclusion and puts forward some key problems and developing trends for future works.更多还原