There is growing interest in how the climate would change under net zero carbon dioxide emissions pathways as many nations aim to reach net zero in coming decades. In today's rapidly warming world, many changes in the climate are detectable, even in the presence of internal variability, but whether climate changes under net zero would be detectable is less well understood. Here, we use a bespoke set of 1000-year-long net zero carbon dioxide emissions simulations branching from different points in the 21st century to examine the detectability of large-scale and local climate changes as time passes under net zero emissions. We find that many changes under net zero become detectable within centuries. While local changes and changes in extremes are more challenging to detect, warming in the Southern Hemisphere and cooling in the Northern Hemisphere becomes detectable at many locations within a few centuries under net zero emissions. We also study detectability of differences in climate indices due to delays in achieving emissions cessation. We find that for global mean surface temperature and other large-scale indices, such as Antarctic and Arctic sea ice extent, the effects of even a five-year delay in emissions cessation are detectable. Short delays in emissions cessation result in significantly different local temperatures for most of the planet, and most of the global population. The long simulations used here help with identifying local climate change signals. Multi-model frameworks will be useful to examine confidence in these changes and ultimately improve understanding of post-net zero climate changes.

A too-weak eddy feedback in models has been proposed to explain the signal-to-noise paradox in seasonal-to-decadal forecasts of the winter Northern Hemisphere. We show that the “eddy feedback parameter’ (EFP) used in previous studies is sensitive to sampling and multidecadal variability. When these uncertainties are accounted for, the EFP diagnosed from CMIP6 historical simulations generally falls within the reanalysis uncertainty. We find the EFP is not independent of the sampled North Atlantic Oscillation (NAO). Within the same dataset, a sample containing larger NAO variability will show a larger EFP, suggesting that the link between eddy feedbacks and the signal-to-noise paradox could be due to sampling effects with the EFP. An alternative measure of eddy feedback, the barotropic energy generation rate, is less sensitive to sampling errors and delineates CMIP6 models that have weak, strong, or unbiased eddy feedbacks, but shows little relation to NAO variability.

Climate models generally project an increase in the winter North Atlantic Oscillation (NAO) index under a future high-emissions scenario, alongside an increase in winter precipitation in northern Europe and a decrease in southern Europe. The extent to which future forced NAO trends are important for European winter precipitation trends and their uncertainty remains unclear. We show using the Multimodel Large Ensemble Archive that the NAO plays a small role in northern European mean winter precipitation projections for 2080-2099. Conversely, half of the model uncertainty in southern European mean winter precipitation projections is potentially reducible through improved understanding of the NAO projections. Extreme positive NAO winters increase in frequency in most models as a consequence of mean NAO changes. These extremes also have more severe future precipitation impacts, largely because of mean precipitation changes. This has implications for future resilience to extreme positive NAO winters, which frequently have severe societal impacts.

Projections of the winter North Atlantic circulation exhibit large spread. Coupled Model Intercomparison Project archives typically provide only a few ensemble members per model, rendering it difficult to quantify reducible model structural uncertainty and irreducible internal variability (IV) in projections. We estimate using the Multi-Model Large Ensemble Archive that model structural differences explain two-thirds of the spread in late 21st century (2080-2099) projections of the winter North Atlantic Oscillation (NAO). This estimate is biased by systematic model errors in the forced NAO response and IV. Across the North Atlantic, the NAO explains a substantial fraction of the spread in mean sea level pressure (MSLP) projections due to IV, except in the central North Atlantic. Conversely, the spread in North Atlantic MSLP projections associated with model differences is largely unexplained by the NAO. Therefore, improving understanding of the NAO alone may not constrain the reducible uncertainty in North Atlantic MSLP projections.

Interactions between ocean basins affect El Niño Southern Oscillation (ENSO), altering its impacts on society. Here, we explore the effect of Atlantic Multidecadal Variability (AMV) on ENSO behaviour using idealized experiments performed with the NCAR-CESM1 model. Comparing warm (AMV+) to cold (AMV-) AMV conditions, we find that ENSO sea surface temperature (SST) anomalies are reduced by ~10% and ENSO precipitation anomalies are shifted to the west during El Niño and east during La Niña. Using the Bjerknes stability index, we attribute the reduction in ENSO variability to a weakened thermocline feedback in boreal autumn. In AMV+, the Walker circulation and trade winds strengthen over the Pacific, increasing the background zonal SST gradient. Those background changes shift ENSO anomalies westwards, with wind stress anomalies more confined to the West. We suggest the changes in ENSO-wind stress decrease the strength of the thermocline feedback in the East, eventually reducing ENSO growth rate.

Climate model emulators are widely used to generate temperature projections for climate scenarios, including in the recent IPCC Sixth Assessment Report. Here we evaluate the performance of a two-layer energy balance model in emulating historical and future temperature projections from CMIP6 models. We find that prediction errors can be large (greater than 0.5oC in a given year) and differ markedly between climate models, forcing scenarios and time periods. Errors arise in emulating the near-surface temperature response to both greenhouse gas and aerosol forcing; in some periods the errors due to these forcings oppose one another, giving the spurious impression of better emulator performance. Time-varying and state-dependent feedbacks may contribute to prediction errors. Close emulations can be produced for a given period but, crucially, this does not guarantee reliable emulations of other scenarios and periods. Therefore, rigorous out-of-sample evaluation is necessary to characterize emulator performance.