考虑随机误差的逐日气温序列集合均一化方法研究

项目名称： 考虑随机误差的逐日气温序列集合均一化方法研究

项目编号： No.41505071

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 天文学、地球科学

项目作者： 李珍

作者单位： 中国科学院大气物理研究所

项目金额： 21万元

中文摘要： 均一化的长期气候序列是气候变化研究的基础，但因台站迁址、仪器变更等非自然因素的影响，致使气候序列存在非均一性，进而影响气候变化趋势的确切估计。而由于气候观测序列中存在的各种随机误差及均一化的统计方法本身具有的不确定性，现有各种均一化方法对于相同气候序列中非均一性的估计存在差异。发展新方法以克服均一化结果的不确定性，系进一步改善现有气候变化研究数据基础所亟需。气温作为一个重要的气候要素，目前我国已发展多套均一化的观测序列集，但对于不同方法所得数据集所呈现的不确定性尚无系统的研究。尤其对如何减小这种不确定性尚缺乏方法学上的创新。本项目借鉴EEMD方法在扰动样本集成分析方面的基本思路，以序列均一化多元分析方法MASH为基础，根据原始观测序列变率特征而加入适量随机扰动，发展基于随机误差的逐日气温序列的多样本集合均一化方法E-MASH。项目将为减小逐日气温序列均一化结果的不确定性构建全新的方法基础。

中文关键词： 均一化逐日气温序列；不确定性；非均一性；集合均一化方法（E-MASH）；随机误差

英文摘要： Long-term homogenous meteorological observation series are essential for assessing climate change, dealing with not only mean climate but climate extremes. However, inhomogeneity is hardly avoided because of non-natural changes such as those in observing location, environs, instrument, and algorithms for calculating climate variables. Inhomogeneities or discontinuities in a climate series affect not only estimates of the mean climate and trend but also those of climate extremes in different ways. Homogenized observational climate series developed by different homogenization methods undoubtedly improved the quality of database of regional climate research. However, because of random errors in climate observations, uncertainties in statistical methods, the inhomogeneities estimated based on different homogenization methods contain uncertainties. How to assess and reduce the uncertainties of homogenization results remains a difficult issue. To overcome or reduce uncertainties in homogenized daily temperature series, the present author is to make an ensemble analysis based on the Multiple Analysis of Series for Homogenization (MASH) method, hence to develop a new method (E-MASH) for homogenizing daily temperature series. A number of sample series will be formed by adding random noises to the candidate and / or reference daily temperature series before synthesizing an ensemble of results of MASH. The new method is expected to substantially reduce uncertainties in the results of homogenization of daily temperature series.

英文关键词： homogenized daily temperature series ;uncertainty;inhomogeneity;ensemble homogenization method(E-MASH);random errors