Quasi-decadal Variability Of Tropical Pacific SST And Its Impact On Tropical Cyclones In The Northwest Pacific

The tropical Pacific atmosphere-ocean system exhibits prominent climate variability on both interannual and decadal timescales.Currently,the knowledge of the interannual El Ni(?)oSouthern Oscillation(ENSO)phenomenon has laid a solid foundation for the operational seasonal forecast.On decadal timescales,however,most previous studies have focused on a few major decadal modes outside the tropics,with less attention paid to the tropical Pacific region.The tropical Pacific decadal variability is an important component of the Pacific decadal variability.So far,its observational features and physical nature have not been well understood,which severely impedes reliable decadal prediction and long-term projection.Based on multiple observational reanalysis datasets,CMIP6 multi-model simulations,dynamical diagnoses,and atmospheric general circulation model experiments,this study explores the observational features,pattern formation mechanisms,and dynamical origin of the tropical Pacific decadal variability over the past seven decades.Its representation in current climate models and regional impacts on western North Pacific tropical cyclone(TC)activities are also examined.The main conclusions are summarized as follows:(1)Over the past seven decades,the Central-Pacific(CP)ENSO-like quasi-decadal variability is the most dominant component of the tropical Pacific decadal variability.The tropical Pacific atmosphere-ocean system exhibits pronounced variability on both interannual and decadal timescales.Being the leading interannual climate mode,the ENSO takes up about60-70% of the total sea surface temperature(SST)variance with its action center located in the equatorial eastern Pacific.In addition,a prominent quasi-decadal(QD)variability can also be detected in the tropical Pacific region.The QD signal is most pronounced in the equatorial central Pacific and the subtropical northeastern Pacific and accounts for about 15-25% of the total SST variance.According to its temporal evolution,the QD variability has strengthened after the 1990 s compared to the early stages.The QD variability is largely originated from the equatorial region and exhibits basin-wide spatial structure reminiscent of the Central Pacific(CP)ENSO/Pacific meridional mode(PMM)based on its canonical phase evolution.The upper-ocean dynamical memory,nevertheless,is very weak for the QD variability around its transition phase,which is very different from the interannual ENSO phenomenon.(2)The tropical Pacific quasi-decadal variability is mainly originated from ENSOinduced nonlinear dynamical cooling while simultaneous large-scale ocean-atmosphere feedbacks are crucial for its pattern formation.Dominated by the thermocline feedback and the three-dimensional advective feedback,the integrated effects of oceanic feedbacks are mostly confined around the equator and contribute positively to local SST anomalies.The combined effects of atmospheric feedbacks play negative and positive roles in SST anomalies over the equatorial central Pacific and subtropical regions and are mainly related to cloudradiation-SST feedback and wind-evaporation-SST feedback,respectively.Being an important precursor,the nonlinear dynamical cooling over the equatorial central to western Pacific induced by strong El Ni(?)o events is proposed as the dynamical origin of the QD variability.During the period from 1970 to 2018 with relatively strong QD variance,strong El Ni(?)o events roughly occur once per decade and generally match the phase transition of the QD variability from positive to negative phases,which can be used to explain the observed dominant QD periodicity.(3)The underestimated physical processes of tropical Pacific quasi-decadal variability in CMIP6 models are largely induced by the common cold bias of equatorial mean SST.Most of the contemporary climate models can reproduce the ENSO-like decadal SST pattern as well as the lead-lag relationship between equatorial and subtropical signals.In particular,the simulation of multi-model averaged Ni(?)o4 SST intensity and the associated ocean-atmosphere feedbacks are consistent with observations.Nevertheless,the mean subtropical northeast Pacific SST intensity is underestimated and only reaches about half of the observed magnitude.This underestimation is possibly related to the systematic SST cold bias over the equatorial cold tongue region and the consequent weak wind-evaporation-SST feedback.Even worse,the subtropical shortwave radiation feedback also exhibits considerable inter-model uncertainties.As most of the models fail to depict the observed SST skewness,the connection between ENSO nonlinearity and tropical Pacific decadal variability is largely underestimated.A small subset of models that reproduce the observed SST skewness can realistically simulate the spatial pattern of ENSO SST,zonal ocean current,and the precursor effect of their combined nonlinear dynamic cooling on tropical Pacific decadal variability.Compared to models with poor skewness performance,those models are characterized by much smaller systematic biases of tropical Pacific mean states.(4)The tropical Pacific quasi-decadal variability exerts prominent modulation effects on western North Pacific tropical cyclone frequency.In the mid-to-late 1990 s,the western North Pacific tropical cyclone(TC)frequency experienced an apparent decadal decrease.It is mainly contributed by the abrupt decrease of TC frequency in the southeastern quadrant of the western North Pacific from July to December,which accounts for 88% of the total change.Local enhanced vertical wind shear and weakened low-level vorticity are two main relevant factors.Further analyses show that the tropical Pacific QD variability can modulate the TC activity in the southeastern quadrant via adjusting these two key environmental factors.Consequently,the QD variability and TC activity exhibit highly consistent decadal changes in extended observational datasets.The QD variability also dominates the statistical relationship between the PMM and TC activity due to its spatial resemblance with the PMM.The significant linkage between the PMM and TC cannot be found after accounting for the QD variability.