Coupled Model Biases Breed Spurious Low‐Frequency Variability in the Tropical Pacific Ocean

• Published in: Geophysical research letters Vol. 45; no. 19; pp. 10,609 - 10,618
• Main Authors: Samanta, Dhrubajyoti, Karnauskas, Kristopher B., Goodkin, Nathalie F., Coats, Sloan, Smerdon, Jason E., Zhang, Lei
• Format: Journal Article
• Language: English
• Published: Washington John Wiley & Sons, Inc 16.10.2018
• Subjects: centennial variability; Climate; Climate change; climate model; Climate models; Climate variability; Climatic analysis; Computer simulation; Cycles; El Nino; El Nino phenomena; El Nino-Southern Oscillation event; ENSO; Evaluation; Future climates; General circulation models; model bias; Ocean currents; Ocean models; Oceans; Periodicity; Predictions; Sea surface; Sea surface temperature; Southern Oscillation; Stochastic models; Stochasticity; Surface temperature; Thermal capacity; Trends; Tropical circulation; Tropical climate; tropical Pacific; Variability; Variance analysis; Wind; Wind fluctuations; Pacific Ocean
• ISSN: 0094-8276; 1944-8007
• DOI: 10.1029/2018GL079455

Coupled general circulation model (GCM) biases in the tropical Pacific are substantial, including a westward extended cold sea surface temperature (SST) bias linked to El Niño–Southern Oscillation (ENSO). Investigation of internal climate variability at centennial timescales using multicentury control integrations of 27 GCMs suggests that a Pacific Centennial Oscillation emerges in GCMs with too strong ENSO variability in the equatorial Pacific, including westward extended SST variability. Using a stochastic model of climate variability (Hasselmann type), we diagnose such centennial SST variance in the western equatorial Pacific. The consistency of a simple stochastic model with complex GCMs suggests that a previously defined Pacific Centennial Oscillation may be driven by biases in high‐frequency ENSO forcing in the western equatorial Pacific. A cautious evaluation of long‐term trends in the tropical Pacific from GCMs is necessary because significant trends in historical and future simulations are possible consequences of biases in simulated internal variability alone. Plain Language Summary The tropical Pacific Ocean exhibits natural climate variability on a wide range of timescales, some of which are similar to the length of instrumental records (~100 years). Characterizing natural cycles with periods of ~100 years is therefore important for detecting and attributing human forced changes. Analysis of climate model simulations shows that a previously defined 100‐year cycle in tropical Pacific sea surface temperatures is the result of mismatches between the models and the real world and therefore may not exist in reality. Our results show that a 100‐year periodicity in the tropical Pacific Ocean is a robust feature of models with an erroneously strong El Niño pattern in the western Pacific Ocean including wind fluctuations. We reveal the causes of 100‐year cycles in the western Pacific, which relies on the ocean's large thermal capacity to smooth out frequent short‐term wind events into slower cycles. Our study highlights the need for cautious interpretations of trends in the tropical Pacific Ocean from climate models due to these possibly spurious 100‐year cycles, especially for attributing historical changes and predicting future climate. If these model mismatches can be corrected, it may allow more accurate predictions of El Niño and long‐term trends over 21st century. Key Points Centennial‐scale variability in the tropical Pacific emerges in some climate models from ocean's integration of higher‐frequency noise Models with El Niño‐Southern Oscillation variance that is too large in the western Pacific exhibit stronger centennial‐scale variability Multimodel ensemble interpretations require a cautious evaluation because of biases in centennial variability in the tropical Pacific