| 摘要: |
| 【目的】分析样本不确定性对干旱评估结果的影响，为防旱减灾工作中干旱的准确评估提供支持。【方法】分别采用参数Bootstrap与非参数Bootstrap抽样方法得到黄河上游（龙羊峡以上区域）1960-2013年共53年的系列标准化降水指数(SPI)值，并通过方差和置信区间宽度2种不确定性量化指标，研究样本不确定性对SPI干旱评估结果的影响。【结果】样本不确定性对SPI评估结果影响显著,其大小随SPI值发生规律性变化，当SPI接近0值时，样本不确定性对SPI评估结果的影响较小；当SPI值减小或增大时，样本不确定性的影响均逐步增大，SPI评估结果的可靠性变差；在降水样本容量相同的条件下，一般基于参数Bootstrap重采样方法的SPI表现为较小的方差和较窄的置信区间宽度。【结论】在应用SPI评估区域干旱状况时，应充分考虑样本不确定性对评估结果的影响；为了减少SPI值的不确定性，宜采用基于非参数Bootstrap的重抽样方法，特别是针对极端事件；若降水样本存在异常值，宜采用基于参数Bootstrap的抽样方法。 |
| 关键词: 干旱评估 样本不确定性 标准化降水指数 Bootstrap重抽样法 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金项目（5167090834） |
|
| Impact of sample uncertainty on SPI index based drought evaluation |
|
ZHANG Ruihan,WANG Yimin,GUO Aijun
|
| Abstract: |
| 【Objective】The impact of sample uncertainty on drought evaluation was analyzed to improve accuracy of drought control and disaster reduction.【Method】The parametric and non parametric Bootstrap methods were adopted to obtain a series of SPI in the upper reaches of the Yellow River (upstream of Longyang Gorge) from 1960 to 2013.The uncertainty quantitative indicators (variance and confidence interval width) were used to investigate the impact of sample uncertainty on SPI index based drought evaluation.【Result】Sample uncertainty had significant impacts on SPI evaluation results.When the SPI was close to 0,sample uncertainty had less effect.Sample uncertainty increased and reliability of SPI evaluation decreased when SPI decreased or increased.With same sample size,the parametric Bootstrap method based SPI had smaller variance and confidence interval width.【Conclusion】The impact of sample uncertainty should be considered when using SPI for draught evaluation.The non parametric Bootstrap resampling method should be used to reduce the uncertainty,especially for extreme events.If there are abnormal values in rainfall samples,the parametric Bootstrap sampling method should be adopted. |
| Key words: drought evaluation sampling uncertainty standardized precipitation idex (SPI) Bootstrap resampling method |
