D. M. Smith, A. A. Scaife, R. Eade, P. Athanasiadis, A. Bellucci, I. Bethke, R. Bilbao, L. F. Borchert, L.-P. Caron, F. Counillon, G. Danabasoglu, T. Delworth, F. J. Doblas-Reyes, N. J. Dunstone, V. Estella-Perez, S. Flavoni, L. Hermanson, N. Keenlyside, V. Kharin, M. Kimoto, W. J. Merryfield, J. Mignot, T. Mochizuki, K. Modali, P.-A. Monerie, W. A. Müller, D. Nicolí, P. Ortega, K. Pankatz, H. Pohlmann, J. Robson, P. Ruggieri, R. Sospedra-Alfonso, D. Swingedouw, Y. Wang, S. Wild, S. Yeager, X. Yang & L. Zhang
Work Package 1
Successful forecasts of North Atlantic atmospheric variability have been hindered by deficiencies in climate models which cause the predictable part of future change relative to random noise to be too small, leading to high uncertainty levels in the forecasts. This study combines a very large number of climate simulations from a range of climate models and applies a post-processing technique that mathematically adjusts for the weak predictable signal. The team found that our current generation of models is then able to produce confident predictions of North Atlantic climate variability decades into the future, greatly improving our ability to predict changes in Atlantic temperatures and European rainfall patterns. The North Atlantic influences various European weather patterns such as storm activity and extreme rainfall, and better predictions of its changes will allow policymakers and those affected by climate change to pre-emptively put measures in place to adapt to and mitigate its possible impacts.
Reducing the amount of uncertainty in our climate model projections is essential for the accurate prediction of future climate change and our efforts to mitigate or adapt to its effects, and a key objective of EUCP. Model simulations agree in some cases, such as large-scale temperature changes, but more complex phenomena like regional precipitation have a large amount of uncertainty in their predictions, hampering our efforts to make reliable projections. A major problem with many predictions is that the predicted change signal is too small in relation to the random noise caused by internal climate variability. This paper seeks to address this issue, focussing on European winters.
The team found that the North Atlantic winter climate is highly predictable and successful decadal forecasts can be made provided a very large number of simulations are available. However, post-processing techniques are required to address current model deficiencies. Crucially, the models severely underestimated the predictable signal of the North Atlantic Oscillation (NAO) an important atmospheric pressure gradient that has a major influence on European climate. This issue can be addressed using mathematical criteria to define a subset of the ensemble based on the NAO predictions. This informed selecting from the larger ensemble leads to a better representation of the magnitude of the NAO variability and related effects, leading to improved predictions of the Atlantic Multi-decadal Variability (AMV; a long-term pattern of oscillations in sea surface temperature) and northern European rainfall, proving that these phenomena can be predicted confidently decades in advance, despite current model deficiencies.
This paper utilises an ensemble of 169 climate simulations from 13 different models. These simulations produced retrospective forecasts of past climate conditions (known as hindcasts) beginning each year from 1960 to 2005, with means analysed for years 2-9 from each start date, focussing on the models’ decadal prediction abilities. These simulations were then compared with observations of temperature, rainfall and sea-level pressure. A post-processing procedure was used that first inflates the variance of the ensemble mean NAO signal to match the variance of the predictable signal in the observations, excluding random noise. A selection process was then applied to the original simulations, choosing the 20 members with a simulated NAO that best matched the variance-adjusted ensemble mean NAO. This subset of the larger ensemble aims to address the incorrect balance of driving forces in climate models and bring out the influence of the NAO.
These findings show that our current climate models can be used to confidently predict North Atlantic climate variability decades in advance. This means we could predict changes in weather types influenced by the North Atlantic, such as storminess or extreme rainfall, allowing policymakers and industries to institute measures in advance to mitigate the impact of changes in North Atlantic climate.
Quantifying signals and uncertainties in climate models is essential for the detection, attribution, prediction and projection of climate change. Although inter-model agreement is high for large-scale temperature signals, dynamical changes in atmospheric circulation are very uncertain. This leads to low confidence in regional projections, especially for precipitation, over the coming decades. The chaotic nature of the climate system may also mean that signal uncertainties are largely irreducible. However, climate projections are difficult to verify until further observations become available. Here we assess retrospective climate model predictions of the past six decades and show that decadal variations in North Atlantic winter climate are highly predictable, despite a lack of agreement between individual model simulations and the poor predictive ability of raw model outputs. Crucially, current models underestimate the predictable signal (the predictable fraction of the total variability) of the North Atlantic Oscillation (the leading mode of variability in North Atlantic atmospheric circulation) by an order of magnitude. Consequently, compared to perfect models, 100 times as many ensemble members are needed in current models to extract this signal, and its effects on the climate are underestimated relative to other factors. To address these limitations, we implement a two-stage post-processing technique. We first adjust the variance of the ensemble-mean North Atlantic Oscillation forecast to match the observed variance of the predictable signal. We then select and use only the ensemble members with a North Atlantic Oscillation sufficiently close to the variance-adjusted ensemble-mean forecast North Atlantic Oscillation. This approach greatly improves decadal predictions of winter climate for Europe and eastern North America. Predictions of Atlantic multidecadal variability are also improved, suggesting that the North Atlantic Oscillation is not driven solely by Atlantic multidecadal variability. Our results highlight the need to understand why the signal-to-noise ratio is too small in current climate models, and the extent to which correcting this model error would reduce uncertainties in regional climate change projections on timescales beyond a decade.