| 摘要: |
| 【目的】研究极值理论(extreme value theory,EVT)法和马尔科夫链(Markov chain,MC)模型在极限干旱历时计算中的预测精度，为区域干旱历时预报提供依据。【方法】以黄河流域兰州、龙门、白马寺3个水文站年径流序列为例，利用EVT法和MC模型2种方法分别计算不同截断水平对应的极限干旱历时(Expected longest duration,E(LT))，并采用纳什效率系数(Nash Sutcliffe efficiency coefficient,NS)和平均绝对百分误差(Mean absolute percentage error,MAPE)评价2种方法的计算精度。【结果】兰州、龙门、白马寺3个水文站采用EVT法计算的极限干旱历时值E(LT)与对应观测值(Observed longest duration,LT-ob)的NS和MAPE分别为0.921 0，0.854 6，0.929 4和0.046 0，0.155 5，0.039 1；而一阶马尔科夫链(one-order Markov chain,MC-1)模型结合hazen绘点位置公式计算的极限干旱历时值与对应观测值的NS和MAPE分别为0.824 7，0.668 1，0.908 1和0.206 3，0.327 3，0.104 6。2种方法均可取得较好的计算效果，但EVT法的计算精度优于MC-1模型。【结论】EVT法较MC-1模型所需要的参数少，计算过程较为简单，且计算结果误差小，是黄河流域可行的极限干旱历时计算方法。 |
| 关键词: 极限水文干旱历时 马尔科夫链 极值理论 径流预测 黄河流域 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金项目（51479171，51179160，50879070）；高等学校博士学科点专项科研基金项目(20110204110017) |
|
| Calculation of critical hydrological drought duration based on EVT method and MC model |
|
ZHAO Wei,SONG Songbai,ZHANG Gengxi
|
| Abstract: |
| 【Objective】This study investigated the accuracy of the extreme value theory (EVT) and the Markov chain (MC) model in predicting critical drought duration on regional scale.【Method】Based on the annual runoff data of 3 hydrological stations in the Yellow River Basin,critical drought duration was modeled using two methods with different threshold levels.The best method was obtained based on Nash Sutcliffe efficiency coefficient and mean absolute percentage error.【Result】The values of NS and MAPE based on EVT at Lanzhou,Longmen and Baimasi hydrological stations were 0.921 0,0.854 6,0.929 4 and 0.046 0,0.155 5,0.039 1 respectively,while those based on the one-order Markov chain model (MC-1) and hazen were 0.824 7,0.668 1,0.908 1 and 0.206 3,0.327 3,0.104 6,respectively.High accuracy was obtained by both methods,and EVT was better.【Conclusion】Compared with the MC-1 models,EVT was simpler,more accurate with less parameters in predicting critical drought durations for the Yellow River Basin. |
| Key words: critical hydrologic drought duration Markov chain extreme value theory runoff forecast Yellow River Basin |
