基于马尔科夫模型的新疆水文气象干旱研究
收稿日期: 2014-12-30
网络出版日期: 2014-09-20
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
Received date: 2014-12-30
Online published: 2014-09-20
Copyright
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)。
孙鹏 , 张强 , 白云岗 , 张江辉 , 邓晓宇 , 刘剑宇 . 基于马尔科夫模型的新疆水文气象干旱研究[J]. 地理研究, 2014 , 33(9) : 1647 -1657 . DOI: 10.11821/dlyj201409006
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.
Fig. 1 Location of hydrological stations, water reservoirs and irrigation areas in Tarim river basin图1 塔里木河流域水文站点、主要水库以及灌区地理位置示意图 |
Tab. 1 SPI-SRI categories表1 SPI-SRI二维联合干旱指标的状态划分 |
| 状态 | SPI值 | SRI值 | 马氏距离D | |
|---|---|---|---|---|
| 状态1 | 气象、水文无旱 | -0.5≤SPI≤0.5 | -0.5≤SPI≤0.5 | D<0.5 |
| 状态2 | 气象、水文湿润 | SPI>0.5 | SRI>0.5 | D≥0.5 |
| 状态3 | 气象干旱、水文湿润 | SPI<-0.5 | SRI>0.5 | D≥0.5 |
| 状态4 | 气象、水文干旱 | SPI<-0.5 | SRI<-0.5 | D≥0.5 |
| 状态5 | 气象湿润、水文干旱 | SPI>0.5 | SRI<-0.5 | D≥0.5 |
(9)
Fig. 2 The SRI and SPI correlation plots for the coupled hydrological and meteorological stations in Tarim river basin图2 塔河流域水文站和气象站对应综合干旱状态的相关关系 |
Fig. 3 Values of AIC criteria for 2 analyzed models of SPI-SRI class time series图3 一阶、二阶马尔科夫链模型的AIC信息准则值 |
Fig. 4 One-step transition probability matrix in Tarim river basin图4 塔河流域各状态的一步转移概率 |
Fig. 5 The return period and the expected residence time of a given SRI-SPI class developed for individual locations in Tarim river basin图5 塔河流域各站点SRI-SPI分级的重现期(A)和期望停留时间(B) |
Fig. 6 The interpretation of the developed first passage of times in terms of hazard of drought for mation, evolution and persistence and drought hazard developed in Tarim river basin图6 塔河流域各站点平均首达时间演绎干旱灾害形成、演化和持续性 |
Fig. 7 Probability to arrive at class 4 and class 5 in 1-6 months starting from class 1, class 2 and class 3图7 预测1-状态3综合干旱状态在未来1-6月的干旱(状态4-状态5)发生概率 |
The authors have declared that no competing interests exist.
| [1] |
|
| [2] |
|
| [3] |
|
| [4] |
|
| [5] |
|
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
| [23] |
|
| [24] |
|
| [25] |
|
/
| 〈 |
|
〉 |