4SEE: Sources and Solutions for Simulated Signal-to-noise Errors and their Effects on climate predictions and projections

Making atmospheric predictions beyond short-term weather forecasts requires predictable signals that change the likelihood of particular climate events amidst the unpredictable atmospheric noise. In seasonal climate prediction, numerical models are used to forecast the evolution of the atmosphere and ocean for the next few months. These models produce forecasts of North Atlantic climate that, on average, better predict the real world than themselves. Given that models and initial conditions are imperfect, we would expect the opposite. This has become known as the signal-to-noise paradox (SNP). The SNP is interpreted as an underestimation of the predictable signal in models, but its root cause remains a mystery. A critical open question is whether the problem also affects multidecadal projections of future climate - the basis for decision-making on climate change - which are derived from similar models to those used in seasonal forecasting. Resolving the SNP would enable better forecasts at critical adaptation planning timescales and is needed to provide credible projections of future climate. 4SEE aims to identify the physical causes of the SNP and develop pathways to eradicate it, through a unique approach examining the SNP across timescales, from subseasonal predictions to multidecadal projections. The potential for common mechanisms that could explain the SNP across timescales motivates our focus on three hypotheses for model errors: too weak transient eddy feedback, insufficient ocean-atmosphere interaction, and weak tropical-extratropical teleconnections. Application of process-level understanding and novel statistical methods will allow us to trace error sources across timescales. We focus on the North Atlantic (NA) region where the SNP is known, but the mechanisms being investigated are likely relevant for other regions. 4SEE has four interconnected objectives, which address all aspects of Highlight Topic C: O1: Advance the detection and diagnosis of the SNP across timescales O2: Identify the causes and nature of the SNP in climate predictions O3: Determine the implications of the SNP for multidecadal projections O4: Provide projections calibrated for the SNP and pathways to eradicate it Novel diagnoses using large ensembles in O1 combined with new high-resolution datasets in O2 will allow us to identify the causes of the SNP and, in turn, assess their implications for projections in O3. O1-O3 will inform the development of calibration techniques in O4. To achieve these objectives, 4SEE will analyse an unprecedented breadth and depth of model simulations, including: ensembles from multiple subseasonal-to-decadal prediction systems and climate models; new large subseasonal and seasonal ensemble hindcasts from the Met Office; and novel predictability and sensitivity experiments at increased atmosphere and ocean resolution with three independent models. Increased model resolution is used as a tool to determine model deficiencies leading to the SNP, and is a potential solution for them. 4SEE unites experts in seasonal-to-decadal prediction, climate modelling and projections, atmosphere-ocean dynamics and interaction, and statistical modelling and calibration. It is motivated by, and builds upon, the team’s extensive work on the SNP: members of the team first uncovered the SNP and have led efforts to quantify, understand, and solve it. The team’s unrivalled expertise and unique multi-timescale approach will ensure successful outcomes – of major relevance to climate prediction, attribution of historical climate change, and projections of future climate. Beneficiaries include meteorological centres to improve forecasts; scientists and policymakers who use climate predictions for research and decision making; and broader end-users of climate services.