Indian Ocean Dipole in CMIP5 and CMIP6: characteristics, biases, and links to ENSO

Abstract

Accurately representing the Indian Ocean Dipole (IOD) is crucial for reliable climate predictions and future projections. However, El Niño-Southern Oscillation (ENSO) and IOD interact, making it necessary to evaluate ENSO and IOD simultaneously. Using the historical simulation from 32 fifth phase of Coupled Model Intercomparison Project (CMIP5) models and 34 CMIP6 models, here we find that there are some modest changes in the basic characteristics of the IOD and ENSO from CMIP5 to CMIP6. Firstly, there is a slight shift in the seasonality of IOD toward an earlier peak in September in CMIP6, from November in CMIP5. Secondly, inter-model spread in the frequency of ENSO and the IOD has reduced in CMIP6 relative to CMIP5. ENSO asymmetry is still underestimated in CMIP6, based on the skewness of the Niño3 index, while the IOD skewness has degraded from CMIP5. Finally, mean state SST biases impact on the strength of the IOD; the Pacific cold tongue mean state is important in CMIP5, but in CMIP6 the Pacific warm pool mean state is more important.

Dynamics for El Niño-La Niña asymmetry constrain equatorial-Pacific warming pattern

“…Figure 2a shows that the standard deviation of detrended Niño-3 SST anomalies (σ ENSO ; 150°-90°W, 5°S-5°N) is very close to the observations on average, however, the multimodel mean of the skewness (γ ENSO ) cannot be statistically distinguished from zero. Thus, CMIP climate models fail badly in simulating ENSO skewness that is about 1 in observations, indicating no improvement in CMIP6 compared to earlier CMIP phases [16][17][18]31 .…”

“…This ENSO asymmetry is known to result in a residual warming signal, rectifying to warmer mean-state ocean temperatures in the eastern equatorial Pacific 4 – 9 . However, ENSO asymmetry is poorly reproduced in most climate models 13 , 16 , 31 . The poor simulation of ENSO skewness also undercuts the models’ ability to simulate a realistic occurrence percentage of extreme El Niño events 32 , potentially reducing the nonlinear rectification effect onto the climate mean state 7 , 16 , 25 , 33 .…”

What Determines the Lagged ENSO Response in the South‐West Indian Ocean?

“…Compared to other climate models, NCAR CESM1 represents well the main features of Indo‐Pacific variability, such as ENSO (Taschetto et al., 2014; Cai et al., 2005), the Indian Ocean Dipole (IOD; McKenna et al., 2020), and Indian and Australian monsoons (Jourdain et al., 2013). Despite overly strong ENSO variability in the model (Figure ), its seasonality is realistically reproduced in the CESM LME, with the Niño3.4 SSTa peak in December (Figure ).…”

Interdecadal modulation of interannual ENSO‐Indian summer monsoon rainfall teleconnections in observations and CMIP6 models: Regional patterns

“…In most CMIP6 models, ENSO peaks in boreal winter except for a group of GISS models but strength is largely differing among the models. The observations show that the IOD peak phase is seen in October, while most models display a slightly early peak in September (Figure 10b–f) consistent with earlies studies (e.g., McKenna et al ., 2020). In the case of the Atlantic Niño, a peak is seen from May to August in observations, whereas most models show a delay by a month.…”