Wengel, Christian (2018) Equatorial Pacific Variability in Climate Models. (PhD/ Doctoral thesis), GEOMAR, Kiel, Germany, 161 pp.

Preview

Text Dissertation_Wengel_Christian.pdf - Published Version Available under License Creative Commons: Attribution 4.0. Download (15MB) | Preview

Abstract

The equatorial Pacific is subject to strong variability on various timescales, which affects climate on a regional as well as on a global scale. The dominant modes of variability are the eastern equatorial Pacific (EEP) sea surface temperature (SST) annual cycle (AC) and the interannual El Niño/Southern Oscillation (ENSO). A realistic simulation of the EEP SST AC and ENSO in climate models together with a profound understanding of the underlying simulated dynamics is crucial for a robust employment of climate models in many different aspects. In this thesis, the simulation of the EEP SST AC and ENSO in climate models is investigated. The EEP SST AC originates from complex interactions of the coupled ocean-atmosphere system and its realistic representation presents a long-standing difficulty in climate models. This thesis investigates the simulation of the EEP SST AC in a set of coupled experiments with the Kiel Climate Model (KCM) which differ in atmospheric resolution. The KCM experiment employing coarse atmospheric resolution depicts significant biases in the EEP SST AC concerning the phase as well as the amplitude of the seasonal variation of the equatorial cold tongue. A large portion of these biases is linked to an erroneous simulation of zonal surface winds, which is associated with an incorrect representation of rainfall to the north and south of the equator. An additional source for EEP SST AC biases originates from a simulated deficit in shortwave radiation related to cloud cover biases. Analyzing the analogous uncoupled atmospheric model integrations forced by observed SSTs suggests that zonal wind and cloud cover biases are inherent to the atmospheric model component. When atmospheric model resolution is enhanced, both wind and cloud cover biases are markedly reduced and the simulation of the EEP SST AC improves. The effect of enhanced atmospheric resolution is, on the one hand, to reduce convection biases over the equatorial Pacific sector and, on the other hand, to improve the simulation of surface winds near landmasses as a result of a refined representation of orography. A subset of models from the 5th phase of the Coupled Model Intercomparison Project (CMIP5) exhibits very similar biases and associated dynamics of the EEP SST AC to those identified in the KCM. The interannual variability associated with ENSO is characterized by a distinct seasonal phase locking with strongest SST anomalies (SSTa) during boreal winter and weakest SSTa in boreal spring. This feature is here investigated in an ensemble of KCM integrations created from perturbed atmospheric physics. The KCM ensemble-mean depicts a realistic seasonal phase locking of the SST variability and of the relevant ENSO feedbacks as inferred from conducting a Bjerknes Stability index analysis. However, the amplitude of the seasonal phase locking is underestimated, which is linked to an excessive simulation of the equatorial cold tongue that reduces the amplitude of the simulated feedbacks. The simulation of eastern equatorial SST variability, mean-state SST and ENSO feedbacks is very sensitive to perturbed atmospheric physics. KCM simulations with a more realistic mean state and ENSO feedbacks also exhibit a more realistic seasonal ENSO phase locking. A similar relationship also is obtained from a set of CMIP5 models. A problematic feature of ENSO simulation in climate models is the large diversity in the simulated strength of ENSO variability. This thesis investigates ENSO-amplitude diversity in a CMIP5 multi-model ensemble by means of the linear recharge oscillator model, which reduces ENSO dynamics to a two-dimensional problem in terms of central and eastern equatorial SSTa (T) and equatorial heat content anomalies (h). Two major sources of the diversity are identified: One originates from stochastic forcing of T and h, the other from interactions of the dynamical processes. The latter suggests competing effects of the growth rate of T and h and the stochastic forcing. These identified sources explain more than 80% of the ENSO-amplitude diversity in the CMIP5 multi-model ensemble.

Actions (login required)

View Item