大气科学  2017, Vol. 41 Issue (4): 752-766 PDF
CMIP5模式对EU、WP遥相关型的模拟评估和预估

1 中国科学技术大学地球和空间科学学院, 合肥 230026
2 甘肃省气象局西北区域气候中心, 兰州 730000

Evaluation and Estimation of Eurasian and West Pacific Teleconnection Pattern in CMIP5
LU Guoyang1,2, REN Baohua1, MA Pengli2, ZHENG Jianqiu1
1 School of Earth and Space Sciences, University of Science and Technology of China, Hefei 230026
2 Lanzhou Regional Climate Center, Gansu Province Meteorology Bureau, Lanzhou 730020
Abstract: The historical and future relationships between Eurasian (EU) and West Pacific (WP) teleconnection patterns, and regional winter temperature and precipitation in East Asia are evaluated by using 14 general circulation models from the Coupled Model Intercomparison Project phase 5. The main conclusions are as follows:(1) Most of the models in CMIP5 have a good ability in simulating the interannual variabilities and spatial patterns of EU and WP, but there still exist certain deviations in the simulated positions of EU and WP. (2) Some of these models could reproduce the negative correlativity between EU and East Asia and Northwest Pacific surface air temperature, but their skill in simulating the relationship between EU and precipitation in North China and Huang-Huai valley was poor. Meanwhile, all of models underestimated the relationship between EU and temperature and precipitation in East Asia. (3) Each individual model and multi-model ensemble mean both showed a high capability in simulating the relationship between WP and temperature in East Asia-Western Pacific region. The positive correlation between WP and precipitation in Northeast Asia and the Okhotsk Sea could be well reproduced; however, their capability for the simulation of the negative correlation from mainland China to western Pacific was poor. It is found that all the models could better simulate the relationship between WP and temperature compared to the simulation of relationship between WP and precipitation. (4) In terms of simulation ability score, CSIRO-Mk3.6.0 on the EU simulation is the strongest and CNRM-CM5 on WP is the best. HadCM3 has the worst model skill scores. (5) Under the RCP4.5 scenario, EU and WP in the future tend to be slightly more in negative phase. Significantly correlated area between EU and temperature in East Asia would shift southeastward based on CSIRO-Mk3.6.0 simulation, while the correlation between EU and precipitation would be insignificant. The correlated area between WP and temperature would shift westward in the high latitudes and eastward in the low latitudes, and the correlation between WP and precipitation would become stronger according to CNRM-CM5 simulation.
Key words: CMIP5      Teleconnection pattern      Regional climate      Model evaluation
1 引言

Wallace and Gutzler（1981）对北半球冬季500 hPa位势高度距平场做单点相关分析时，发现存在5种遥相关型分布：太平洋—北美型（PNA）、东大西洋型（EA）、西大西洋型（WA）、西太平洋型（WP）以及欧亚遥相关型（EU），其中WP和EU型遥相关对东亚冬季气候有显著的影响。EU存在三个异常中心，呈纬向分布特征，分别位于斯堪的纳维亚、乌拉尔地区以及日本上空；当斯堪的纳维亚和日本上空出现负异常高度场，乌拉尔地区为正位势高度，则为正位相EU。WP型主要特征为西北太平洋地区和鄂霍次克海域位势高度场的跷跷板变化，当正位相WP时，西北太平洋为正异常位势高度，日本上空急流减弱，北部鄂霍次克海为负异常位势高度，阿留申低压偏低。

EU和WP对我国以及东亚地区冬季气候有着非常重要的影响，特别是气温和降水。施能（1996）认为北半球遥相关年代际变化是中国冬季气候变化的重要原因之一；李维京和丑纪范（1990）指出冬季欧亚遥相关是长江中下游冬季降水异常的主要因子；刘毓赟和陈文（2012）认为EU正位相时，东亚冬季风偏强，从而使我国东部降温、降水减少。关于WP型遥相关研究中，吴洪宝（1993）分析指出西太平洋型与我国东部地面气温同期相关显著，且12月西太平洋型与1月中国东部气温非同期相关显著；李勇等（2007）指出冬季WP遥相关与我国冬季和降水存在显著的大范围正相关：高WP年，西伯利亚高压强度偏弱，东亚冬季风减弱，我国气温偏高，降水偏多。

2 数据和方法 2.1 资料

2.2 研究方法

EU、WP指数（EUI、WPI）定义如下：

 $\begin{array}{l} {\rm{EUI}} = - 0.25 \times {Z^{\rm{*}}}(55^\circ {\rm{N, }}20^\circ {\rm{E}}) + 0.5 \times \\ {Z^{\rm{*}}}(55^\circ {\rm{N, }}75^\circ {\rm{E}}) - 0.25 \times {Z^{\rm{*}}}(40^\circ {\rm{N}},{\rm{ }}145^\circ {\rm{E}}){\rm{ ,}} \end{array}$ (1)
 ${\rm{WPI}} = 0.5 \times [{Z^{\rm{*}}}\left( {60^\circ {\rm{N}},{\rm{ }}155^\circ {\rm{E}}} \right) - {Z^{\rm{*}}}\left( {30^\circ {\rm{N, }}155^\circ {\rm{E}}} \right)]{\rm{ }},$ (2)

 $S = \frac{{4{{(1 + R)}^4}}}{{{{({{\hat \sigma }_f} + \frac{1}{{{{\hat \sigma }_f}}})}^2}{{(1 + {R_0})}^4}}},$ (3)

3 对EU、WP遥相关年际变化模拟

4 空间型分布模拟效果 4.1 对EU空间型分布模拟效果

 图 1 观测和15个模式资料的北半球欧亚地区冬季500 hPa位势高度距平场EOF分解所得EU型分布 Figure 1 Spatial patterns of EU based on EOF analysis of the H500 (500-hPa geopotential height) anomalies from observations (lower right corner) and simulations of 15 GCMs

 图 2 Intermodel-EOF分解EU型（a, b）前两个空间模态以及（c, d）PC（模态对应的时间系数）序列 Figure 2 (a) The first and (b) second Intermodel-EOF modes of EU pattern simulated by 14 GCMs; (c, d) corresponding two PCs (Principal Components)
4.2 对WP空间型分布模拟效果

NCEP再分析资料显示，东亚及西太平洋地区冬季500 hPa位势高度场EOF分解第一模态为南北偶极子空间分布，异常中心分别位于西太平洋—日本上空和鄂霍次克海东北部区域，其时间序列解释方差可达31.1%（如图 3p）。所选CMIP5模式都能很好的模拟南北偶极子分布形态，并且对西太平洋—日本区的异常中心有较好模拟，但对于鄂霍次克海附近的活动中心模拟效果较差，多数模拟呈现偏北偏西的分布特征，如BCC-CSM1.1（图 3a）、CanESM2（图 3b）、CCSM4（图 3c）、CSIRO-Mk3.6.0（图 3e）、GFDL-CM3（图 3g）、GISS-E2-R（图 3h）、IPSL-CM5A-LR（图 3k）、NorESM1-M（图 3n）、MME（图 3o）。除了模式MRI-CGCM3（EOF3）和MIROC5（EOF2）外，其他模式EOF第一模态都表现为WP型。各模式与再分析资料EOF模态的空间相关系数最高为INM-CM4，可达0.91，其次为CNRM-CM5为0.88。总体对WP空间分布型模拟效果不错，除了HadCM3，其他模式与观测的空间相关均大于0.6。

 图 3 观测和15个模式资料的冬季500 hPa位势高度距平场EOF分解所得WP型分布 Figure 3 Spatial patterns of WP based on EOF analysis of the H500 anomalies from observations (lower right corner) and simulations of 15 GCMs

Intermodel-EOF分析结果如图 4所示。EOF分解第一模态（图 4a）在中高纬度西太平洋上存在一个南北向偶极子分布，除异常中心略向东移外，其与WP型分布相似（图 3p），对应时间序列解释方差可达57.3%，该模态说明模式与观测的WP型的差异主模态为模拟强度差异，其中CNRM-CM5和CSIRO-Mk3.6.0对WP强度模拟效果较好（如图 4c）；第二模态EOF2（图 4b）中，呈现的是东亚大陆和西北太平洋东西向偶极子分布，该模态解释方差为16.5%，说明了对WP型模拟的位置偏差，由PC2可知（图 4d），CCSM4模拟的WP型的位置误差最小，模拟效果较好，而MIROC5和GISS-E2-R存在较大的位置偏差。

 图 4 同图 2，但为Intermodel-EOF分解WP型 Figure 4 Same as Fig. 2, but for Intermodel-EOF modes of WP pattern
5 遥相关与局地气温、降水的关系

5.1 EU与东亚—西太平洋区气温、降水

 图 5 观测和模式资料中EU模态对应的PC序列与冬季东亚地区地表气温相关系数分布（等值线间隔为0.2，粗实线为零线，虚线为负值；浅、深黄（蓝）阴影区为通过95%、99%信度检验） Figure 5 Observed and GCMs simulated correlation coefficients between the PC that corresponds to EU mode and regional winter surface temperature over East Asia-the West Pacific. (Contour interval is 0.2. The thick solid lines indicate the zero contour and the dashed lines indicate negative values. The light and dark yellow (blue) shadings are for values that exceed the 95% and 99% confidence levels, respectively)

 图 6 同图 5，但为PC序列与冬季东亚地区降水量相关系数分布 Figure 6 Same as Fig. 5, but for correlations between the PC and regional winter precipitation

 图 9 模式模拟（a）EU、（b）WP指数与局地表面气温（SAT，红圈）、降水量（Precip，蓝三角）相关性和观测资料的泰勒图（简言之，离REF点的距离越近，说明该模式模拟能力越佳） Figure 9 Taylor diagrams of the relationships of (a) EU and (b) WP indexes with regional surface air temperature (SAT, red circles) and precipitation (Precip, blue triangles)
5.2 WP与东亚—西太平洋区气温、降水

 图 7 同图 5，但为WP指数与冬季东亚地区地表气温相关系数分布 Figure 7 Same as Fig. 5, but for correlations between the WP index and regional winter surface temperature

 图 8 同图 5，但为WP指数与冬季东亚地区降水量相关系数分布 Figure 8 Same as Fig. 5, but for correlations between the WP index and regional winter precipitation

 图 10 各模式对EU、WP遥相关整体模拟能力评分 Figure 10 The skill scores of each model for overall simulation capability of EU and WP
6 RCP4.5情景预估

 图 11 RCP4.5情景下未来标准化EU、WP指数的时间变化（蓝色实线为9年滑动平均，红色虚线为线性趋势线） Figure 11 The future changes in normalized EU and WP indexes under the RCP4.5 scenario (the blue solid line shows the 9-year running mean, the red dashed line indicates the linear trend)

 图 12 RCP4.5情景下CSIRO-Mk3.6.0模式中EU指数与冬季东亚地区（a）气温和（b）降水量的相关分布图（等值线间隔为0.1，粗实线为零线，实线为正，虚线为负，浅、深黄（蓝）阴影表示通过95%、99%信度检验） Figure 12 CSIRO-Mk3.6.0 simulated correlation coefficients between EU index and (a) regional winter air temperature, (b) winter precipitation over East Asia-the West Pacific (contour interval is 0.1. The thick solid lines indicate the zero contour, the solid lines and dashed lines indicate positive and negative values, respectively. The light and dark yellow (blue) shadings are for values that exceed the 95% and 99% confidence levels, respectively)

 图 13 同图 12，但为CNRM-CM5模式中WP指数与冬季东亚（a）气温与（b）降水量相关 Figure 13 Same as Fig. 12, but for the CNRM-CM5 simulated correlation between WP index and regional winter air temperature and precipitation
7 结论

（1）均方差分析可知模式对EU、WP信号的整体年际变率有一定模拟技巧，但对具体时间变化模拟较差。对其空间模态特征模拟效果较好，能很好的再现遥相关的异常中心，但也存在一定的强度和位置偏差。

（2）多数模式能再现EU与东亚以及西北太平洋地区表面气温的负相关性，但显著性区域存在较大差异。多模式集合MME对EU与东亚表面气温的相关性模拟存在低估现象；各模式对EU与我国华北以及黄淮流域地区的负相关的模拟效果存在较大偏差。泰勒图说明模式对EU与气温关系模拟能力高于其与之降水，但大多数模式低估了EU与东亚地区气温、降水的关系。