张芳,董敏,吴统文. 2014. CMIP5模式对ENSO现象的模拟能力评估[J]. 气象学报, 72(1):30-48, doi:10.11676/qxxb2014.011
CMIP5模式对ENSO现象的模拟能力评估
Evaluation of the ENSO features simulations as done by the CMIP5 models
投稿时间:2013-05-15  修订日期:2013-10-22
DOI:10.11676/qxxb2014.011
中文关键词:  CMIP5模式  ENSO  模拟  评估
英文关键词:CMIP5 Models  ENSO  Simulation  Evaluation
基金项目:国家重点基础研究发展计划项目(2010CB951902);公益性行业科研专项(201306048、201006024)。
作者单位E-mail
张芳 国家气候中心, 北京, 100081  
董敏 国家气候中心, 北京, 100081 dongm@cma.gov.cn 
吴统文 国家气候中心, 北京, 100081  
摘要点击次数: 4414
全文下载次数: 12100
中文摘要:
      针对参与耦合模式比较计划(CMIP5)的17个海-气耦合模式对20世纪气候的模拟结果,从热带太平洋海表温度和大气海平面气压变化的综合分析角度较详细评估了模式对厄尔尼诺-南方涛动(ENSO)现象的模拟能力。结果表明,这些模式基本上能模拟出ENSO现象的一些主要特征,包括热带太平洋海温的空间分布及其时空演变特征、与海平面气压变化的关联、ENSO周期变化及锁相特征等,但不同模式的模拟结果仍然差异较大。(1)从模拟的热带太平洋年平均海温的偏差来看,多模式集合平均值与观测的均方根误差小于1.0℃,但单个模式的误差相对要大一些。误差较小的为1.2—1.3℃,多数模式在1.6℃以下,但也有个别模式的误差超过2.0℃。(2)从经验正交函数分解结果来看,热带太平洋实测月平均海表温度距平和海平面气压距平的年际尺度变化第1模态主要表现为ENSO变化特征,第2模态反映的是海温的长期变化趋势。只有少数几个CMIP5模式能够再现这种特征,多数模式所模拟的海温距平/海平面气压距平时空变化的第1、第2特征向量分布顺序与观测分析正好相反,ENSO变成了第2模态,趋势成了最主要的模态。尽管如此,所有模式都能模拟出南方涛动变化与热带太平洋海温距平时空变化的密切关联,无论是作为第1还是第2特征模态,所有模式模拟的南方涛动与热带太平洋海温距平时空变化都有密切相关。(3)谱分析结果表明,ENSO现象具有2—7年的周期,其中,4年的周期最明显。大多数模式模拟的ENSO周期在此范围内,但有些模式的主要周期偏短,为2年左右。个别模式的ENSO主要周期为11年,已超出2—7年的范围。(4)多数模式模拟的厄尔尼诺及拉尼娜的峰值出现在冬季(11—2月),与观测基本吻合。另有少数模式模拟的峰值出现在9—10月,比观测略提前。只有个别模式模拟的峰值出现在夏季,与观测相差太大。
英文摘要:
      The ability of the 17 CMIP5 models in simulating the ENSO phenomenon is examined by using the outputs of these models from the historical experiments of the 20th century. In general, the models can simulate some major characteristics of the ENSO phenomena, such as the mean sea surface temperature (SST) in the tropical Pacific; the temporal and spatial evolution of the SST anomalies; the interactive relation between oceans and the atmosphere; the periodicity of the ENSO; the phase locking feature of the ENSO and so on. There is large difference in the ability of simulating the ENSO between various models. (1) The simulated SST still has some errors in various degrees. This error is small for the multiple model ensemble with the root mean square error (RMSE) between the simulated and observed SST being below 1.0℃ and otherwise for each single model in which the RMSE is larger than this. Some good models can have error of 1.2-1.3℃, majority of the models has errors below 1.6℃, and there are still few models which have RMSE exceeding 2.0℃. (2) According to the Empirical Othorgnal Function (EOF) analyses, the temporal and spatial variation of the simulated SST anomalies and Sea Level Pressure (SLP) anomalies for a few of the models is close to the observation, its first mode is ENSO mode and the corresponding time coefficient represents the ENSO evolution. Its second mode represents the increasing trend of the SST anomaly during the last period of more than 50 years. For most of the models the sequence of the temporal and spatial variation mode of the simulated SST/SLP anomaly is opposite to that of the observation. The increasing trend becomes the first mode with the major variance contribution, while the ENSO becomes the second mode. This means that the mechanism which produces the temperature increase from the CO2 induced greenhouse effect is too strong in these models, while the ENSO oscillation mechanism is rather weak. It is showed that no matter it is the first mode or second mode, the corresponding time coefficient of the southern oscillation has the good correlation with the SST anomaly. This means the CMIP5 models can well represent the close relationship between the El Nio-La Nia and the southern oscillation. (3) The spectral analysis shows that the ENSO phenomenon has 2-7 year quasi-periodicity and the 4 year periodicity is the most obvious. In most of the CMIP5 models the ENSO has the periods of 2-7 years, this is consistent to the observation. But some models have ENSO period of 2 year or so, and few has too long period of 11 years. And, (4) in the simulations of most of models the peak phase of the El Nio/La Nia appears in later fall through winter (November-February), which is consistent to observation. There are also few models whose simulated ENSO peak appears in September-October or even in summer and this is not consistent with observation.
HTML   查看全文   查看/发表评论  下载PDF阅读器
分享按钮