CMIP6模式对西北太平洋异常反气旋的模拟偏差:主要模态及其来源分析

Dominant modes and sources of CMIP6 model biases in simulating the western North Pacific anomalous anticyclone

摘要:

当El Niño发生时,西北太平洋会出现一个异常反气旋(western North Pacific anomalous anticyclone,WNPAC),对东亚气候产生重要影响。本文评估了第六次耦合模式比较计划(Coupled Model Intercomparison Project Phase 6,CMIP6)中的模式对该WNPAC的模拟情况,包括CMIP6模式集合平均以及模式间离散性的主要特征。分析表明,CMIP6模式集合平均模拟的WNPAC位置偏西,且强度偏弱。利用正交经验函数分解出模式间离散性的主模态,其第一模态表现为WNPAC的纬向位置差异,第二模态则为WNPAC强度的差异。前者与El Niño时赤道中东太平洋暖海温异常西伸有关,而最终可以归因到赤道太平洋气候态冷舌的模拟偏差;而后者则与耦合模式中的“风-海温”正反馈强度有关。大部分模式模拟的赤道太平洋气候态SST冷舌偏强,且对“风-海温”正反馈模拟偏弱,使得CMIP6模式整体上模拟的WNPAC偏西且偏弱。

Abstract:

The western North Pacific anomalous anticyclone (WNPAC) is a key atmospheric circulation system linking El Niño-Southern Oscillation (ENSO) events to climate variability in East Asia.During El Niño winters,anomalous warming in the equatorial central-eastern Pacific induces a Gill-type response that generates an anomalous anticyclone over the western North Pacific,exerting far-reaching influences on the East Asian monsoon,winter temperature,and precipitation patterns.Despite its importance,the ability of state-of-the-art climate models to accurately simulate the WNPAC remains insufficiently understood.
This study provides a systematic assessment of 25 models participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) in simulating the climatological WNPAC during the historical period (1940—2014) and further investigates the physical origins of model biases.The results show that the multi-model mean (MMM) significantly underestimates the strength of the WNPAC and shifts its center westward relative to observations.While observations place the anticyclonic center over the Philippine Sea,the CMIP6 MMM locates it over the South China Sea with a substantially weakened amplitude.This systematic bias suggests that CMIP6 models tend to underestimate ENSO teleconnections affecting the East Asian winter climate.
To examine intermodel diversity,empirical orthogonal function (EOF) analysis is applied to the WNPAC in both CMIP6 models and observations.Two leading modes are identified.EOF1 (40.6％ of variance explained) represents zonal displacement of the WNPAC,distinguishing models with westward-versus eastward-shifted anticyclones.Most models exhibit positive principal component (PC1) values relative to observations,indicating a systematic westward bias.EOF2 (25.2％ of variance explained) characterizes variations in WNPAC intensity,with most models underestimating its strength.
Regression analyses based on the principal components reveal the physical mechanisms underlying these biases.The zonal displacement bias (EOF1-type) is strongly associated with the westward extension of El Niño-induced SST anomalies in the central Pacific.Models with stronger equatorial cold tongue biases exhibit enhanced zonal SST gradients and stronger nonlinear zonal advective feedbacks,which shift anomalous convection and the associated Rossby wave response westward.This indicates that an unrealistic mean-state cold tongue is a primary driver of the systematic westward bias of the WNPAC in CMIP6 models.
The intensity bias (EOF2-type) is linked to variations in the strength of the Bjerknes “wind-SST” positive feedback.Models with stronger air-sea coupling simulate greater weakening of the trade winds during El Niño,leading to enhanced SST cooling in the western Pacific and a stronger WNPAC response.The PC2 values show a significant correlation (r=0.65) with independently diagnosed wind-SST feedback strength,confirming that underestimated air-sea coupling in most CMIP6 models contributes to the widespread weak WNPAC bias.
Finally,a skill score is constructed based on the distance between each model’s (PC1,PC2) pair and the observed reference point,weighted by the explained variance.The best-performing models (KIOST-ESM,BCC-ESM1,CAMS-CSM1-0,MPI-ESM1-2-LR,and EC-Earth-Veg) successfully reproduce the location and intensity of the WNPAC,whereas models such as NESM3,CMCC-CM2-SR5,TaiESM,FIO-ESM-2-0,and NorCPM1 exhibit the poorest performance.
Overall,this study demonstrates that systematic biases in CMIP6 models—particularly an overly strong equatorial cold tongue and a weakened wind-SST coupling—jointly lead to a westward-shifted and weakened WNPAC response to El Niño.These findings highlight the uncertainties in simulating ENSO-WNPAC-East Asia teleconnection under future climate change.