Search

The Quasi‐Biennial Oscillation (QBO) is the dominant mode of interannual variability in the tropical stratosphere, with easterly and westerly zonal wind regimes alternating over a period of about 28 months. It appears to influence the Northern Hemisphere winter stratospheric polar vortex and atmospheric circulation near the Earth's surface. However, the short observational record makes unequivocal identification of these surface connections challenging. To overcome this, we use a multicentury control simulation of a climate model with a realistic, spontaneously generated QBO to examine teleconnections with extratropical winter surface pressure patterns. Using a 30‐hPa index of the QBO, we demonstrate that the observed teleconnection with the Arctic Oscillation (AO) is likely to be real, and a teleconnection with the North Atlantic Oscillation (NAO) is probable, but not certain. Simulated QBO‐AO teleconnections are robust, but appear weaker than in observations. Despite this, inconsistency with the observational record cannot be formally demonstrated. To assess the robustness of our results, we use an alternative measure of the QBO, which selects QBO phases with westerly or easterly winds extending over a wider range of altitudes than phases selected by the single‐level index. We find increased strength and significance for both the AO and NAO responses, and better reproduction of the observed surface teleconnection patterns. Further, this QBO metric reveals that the simulated AO response is indeed likely to be weaker than observed. We conclude that the QBO can potentially provide another source of skill for Northern Hemisphere winter prediction, if its surface teleconnections can be accurately simulated. QBO teleconnection with the winter Arctic Oscillation is likely to be real, but observations only weakly constrain its strengthAnalysis using new deep QBO phase measure increases the strength and significance of the winter surface teleconnectionsTeleconnections between the deep QBO and Arctic Oscillation simulated by the HadGEM3 climate model are likely weaker than those observed

Tropical stratospheric variability is investigated in ensembles of historical simulations performed by nine models for the sixth phase of the Coupled Model Intercomparison Project (CMIP6). Realizations from all the models feature a reasonable quasi‐biennial oscillation (QBO). Variability in the zonal mean zonal winds at the equator is found to be coherent among realizations when a model prescribes the CMIP6 ozone forcing. No such coherence is found when a model simulates ozone. The coherence results in an ensemble mean QBO signal with amplitude, depending on the model, 50%–80% of the mean QBO amplitude from a single realization from the same model. The ensemble mean signal is due to synchronization of QBO phases, attributed to the CMIP6 protocol including a QBO signal in the ozone forcing. Coherence in models using the CMIP6 ozone is, therefore, artificial and individual realizations from these models are not completely independent. High above the equator winds in the stratosphere at altitudes from ∼16–40 km repeatedly switch direction from eastward to westward, and back again, roughly every 14 months. This so‐called quasi‐biennial oscillation (QBO) featured in only a subset of the simulations of the historical period that provided input for the latest international assessments of climate change. Variability in the stratospheric winds at the equator in these simulations is found to be remarkably coherent across ensembles of realizations from the individual models, but only for those models that did not simulate ozone variability. Instead, these models followed a common protocol that specified prescribing an ozone distribution that included a QBO in the tropical stratosphere. For each of the models this causes the simulated QBOs in the winds to synchronize across many realizations of the historical period. Consequently variability in the equatorial stratosphere in these realizations becomes coherent. However, this coherence is not real as it is merely an artifact of the simulations following the agreed ozone protocol. Likewise, due to the design of the protocol, individual realizations from these models can not be considered as been entirely independent in the tropical stratosphere. Coupled Model Intercomparison Project phase 6 (CMIP6) historical simulations from several models feature a realistic quasi‐biennial oscillation (QBO) in the equatorial stratosphereQBO phases synchronize across ensembles of realizations from models not able to simulate evolving ozoneThe phase synchronization is artificial and attributed to the CMIP6 protocol prescribing a QBO signal in the ozone forcing data Coupled Model Intercomparison Project phase 6 (CMIP6) historical simulations from several models feature a realistic quasi‐biennial oscillation (QBO) in the equatorial stratosphere QBO phases synchronize across ensembles of realizations from models not able to simulate evolving ozone The phase synchronization is artificial and attributed to the CMIP6 protocol prescribing a QBO signal in the ozone forcing data

The relative importance of anthropogenic aerosol in decadal variations of historical climate is uncertain, largely due to uncertainty in aerosol radiative forcing. We analyze a novel large ensemble of simulations with HadGEM3‐GC3.1 for 1850–2014, where anthropogenic aerosol and precursor emissions are scaled to sample a wide range of historical aerosol radiative forcing with present‐day values ranging from –0.38 to –1.50 Wm–2. Five ensemble members are run for each of five aerosol scaling factors. Decadal variations in surface temperatures are strongly sensitive to aerosol forcing, particularly between 1950 and 1980. Post‐1980, trends are dominated by greenhouse gas forcing, with much lower sensitivity to aerosol emission differences. Most realizations with aerosol forcing more negative than about –1 Wm–2simulate stronger cooling trends in the mid‐20th century compared with observations, while the simulated warming post‐1980 always exceeds observed warming, likelydue to a warm bias in the transient climate response in HadGEM3‐GC3.1. Anthropogenic aerosols have an overall cooling effect on climate due to their interaction with incoming solar radiation and influence on cloud properties. Their emissions have offset some of the historical warming induced by increasing greenhouse gases. However, the magnitude of the cooling induced by anthropogenic aerosol remains poorly constrained. In this study, we use a state‐of‐the‐art climate model, HadGEM3‐GC3.1, driven by different levels of aerosol emissions. This experimental setup tests the sensitivity of simulated historical temperatures to the strength of aerosol forcing in a climate model, all other factors remaining equal. Our results show that the period from 1951 to 1980 is particularly sensitive to aerosol forcing, coinciding with a period of rapid increases in global aerosol emissions and observed cooling over many regions, while temperature trends from 1980 onwards are primarily driven by increases in greenhouse gas concentrations. The observed temperatures over 1951–1980 are best reproduced by simulations with lower aerosol emissions than the standard configuration, implying that this model responds too strongly to aerosol forcing. Concurrently, the simulated temperatures warm faster than observed temperatures from 1980 onwards, suggesting that this climate model also responds more strongly to greenhouse gas forcing than observations suggest. Simulations sampling aerosol forcing uncertainty suggest that aerosol forcing in the HadGEM3‐GC3.1 CMIP6 historical simulations is too largeComparing two key periods (1950–1980 and 1981–2010) with observations suggests a positive bias in the transient climate response in HadGEM3Most realizations with aerosol forcing more negative than about –1 Wm–2cool too much in the mid‐20th century