地理研究 ›› 2014, Vol. 33 ›› Issue (9): 1647-1657.doi: 10.11821/dlyj201409006

• • 上一篇    下一篇

基于马尔科夫模型的新疆水文气象干旱研究

孙鹏1,2,3, 张强1,2,3,**(), 白云岗4, 张江辉4, 邓晓宇1,2,3, 刘剑宇1,2,3   

  1. 1 中山大学 水资源与环境系,广州 510275
    2中山大学 地理科学与规划学院,广东省城市化与地理环境空间模拟重点实验室,广州510275
    3 中山大学 华南地区水循环与水安全广东省普通高校重点实验室,广州 510275
    4 新疆水利水电科学研究院, 乌鲁木齐830049
  • 收稿日期:2014-12-30 出版日期:2014-09-20 发布日期:2014-11-10
  • 通讯作者: 张强 E-mail:zhangq68@mail.sysu.edu.cn
  • 作者简介:

    作者简介:王贺封(1980- ),男,江西抚州人,博士生,讲师,主要研究方向为城市地理信息系统。E-mail:wanghefeng1980@163.com

  • 基金资助:
    新疆维吾尔自治区科技计划项目(201331104);中山大学博士研究生创新人才培养资助项目;国家自然基金项目(41071020)

Transitional behaviors of hydrometeorological droughts in Xinjiang using the Markov chain model

Peng SUN1,2,3, Qiang ZHANG1,2,3(), Yungang BAI4, Jianghui ZHANG4, Xiaoyu DENG1,2,3, Jianyu LIU1,2,3   

  1. 1. Department of Water Resources and Environment, Sun Yat-sen University, Guangzhou 510275, China
    2. School of Geography and Planning, Guangdong Key Laboratory for Urbanization and Geo-simulation, Sun Yat-sen University, Guangzhou 510275, China
    3. Key Laboratory of Water Cycle and Water Security in Southern China of Guangdong High Education Institute, Sun Yat-sen University, Guangzhou 510275, China
    4. Xinjiang Research Institute of Water Resources and Hydropower, Urumqi 830049, China
  • Received:2014-12-30 Online:2014-09-20 Published:2014-11-10

摘要:

在气象干旱SPI和水文干旱SRI的二维变量干旱状态的研究基础上,通过一阶马尔科夫链模型对二维变量干旱状态进行频率、重现期和历时分析,并预测未来非水文干旱到水文干旱的概率,研究结果表明:(1)开都河、和田河在干旱形成中危害大,阿克苏河在干旱演变中危害大,开都河和叶尔羌河在干旱持续中危害大。开都河和叶尔羌河主要以气象水文干旱为主,和田河和阿克苏河以水文干旱为主。(2)开都河连续湿润或者干旱的概率最大,状态2与状态4、状态5的相互转移概率低,和田河和开都河状态4不能一步转移到状态2。(3)在长期干旱预测中,塔河流域从状态2达到状态4或者状态5的概率最低,开都河(或和田河)从非水文干旱状态到状态4的概率最大(或最小),从非水文干旱状态到状态5的概率最小(或最大)。

关键词: 气象水文干旱状态, 干旱预测, 马尔科夫模型, 塔里木河流域

Abstract:

Bivariate SPI-SRI drought index was proposed to describe drought behaviors of the Tarim river basin. Five Meteor-hydrological drought conditions were identified with three drought hazard: meteorological dry, both meteorological and hydrological dry, and hydrological dry conditions. Drought hazard was investigated in terms of drought formation process with the atmospheric and hydrological phases. Stochastic analysis of the developed indicator allows assessing the dynamics of the transition between drought phases. Time series of the meteor-hydrological drought conditions were investigated as discrete state, discrete-time homogenous Markov chain. Analysis of the properties of Markov chain aimed to evaluate probably of transition between different conditions, frequency of each conditions, residence time in each conditions, time required to move from one condition to another, and predict drought hazard in next month. The results indicate that: (1) The droughts have the significant impacts during its development stage in the Kaidu and Hotan river basins. The negative influences of droughts are evident during its evolutions periods in Akesu river basin and during its successive periods in Kaidu and Yarkant river basins. The Kaidu and Yarkant River basins are dominated by meteor-hydrological droughts, and hydrological droughts in the Hotan and Akesu river basins; (2) Occurrence probability is the largest for consecutive wet or drought conditions in the Kaidu river basins and the probability of condition shifts between conditions 2, 4 and 5 in the Kaidu river basin. No drought condition shifts can be expected between conditions 4 and 2 in the Hotan and Kaidu river basins; (3) Forecasting practice of drought conditions, the occurrence probability of drought condition shifts between conditions 2 to 4 and 2 to 5 is the lowest. Largest occurrence probability can be expected for drought condition shifts between non-hydrological drought conditions and condition 4 and the smallest probability could be expected for drought condition shifts between non-hydrological drought and condition 5.

Key words: meteor-hydrological drought conditions, probabilistic prediction, Markov model, Tarim river basin

PACS: 

  • P954