Robust skill of decadal climate predictions

Abstract

There is a growing need for skilful predictions of climate up to a decade ahead. Decadal climate predictions show high skill for surface temperature, but confidence in forecasts of precipitation and atmospheric circulation is much lower. Recent advances in seasonal and annual prediction show that the signal-to-noise ratio can be too small in climate models, requiring a very large ensemble to extract the predictable signal. Here, we reassess decadal prediction skill using a much larger ensemble than previously available, and reveal significant skill for precipitation over land and atmospheric circulation, in addition to surface temperature. We further propose a more powerful approach than used previously to evaluate the benefit of initialisation with observations, improving our understanding of the sources of skill. Our results show that decadal climate is more predictable than previously thought and will aid society to prepare for, and adapt to, ongoing climate variability and change.npj Climate and Atmospheric Science (2019)2:13 ; https://doi.

Understanding the Signal‐to‐Noise Paradox with a Simple Markov Model

“…There is a significant correlation between the potential predictability variance fraction and the ratio of the persistence and noise variance, with correlation coefficients of 0.94 and 0.42 for seasonal and annual NAO predictability, respectively. This high correlation, especially for seasonal NAO index, indicates that the signal‐to‐noise paradox may also exist at seasonal‐to‐annual and longer timescales (e.g., Smith et al, ), as it is notable that the NAO predictability for the observational estimates presents higher values than most of the CMIP5 model simulations. Similar results are also found in other observational datasets such as the ERA20C data.…”

“…Uncertainty in the initial condition will evolve with time, lead to uncertainty of the forecast, and inherently limit the prediction skill (Shukla, 1998;Slingo & Palmer, 2011;Zhang & Kirtman, 2019). The idea of ensemble forecasting from different initial conditions, therefore, has been applied to the operational systems for numerical weather prediction (e.g., Buizza et al, 1999;Hamill et al, 2013;Murphy, 1988) and subsequently seasonal-to-decadal climate prediction (e.g., Kirtman et al, 2014;MacLachlan et al, 2015;Smith et al, 2019). Ensemble forecasting can quantify the uncertainty due to imperfect initial states and model formulations and has become routinely used in practice.…”

Multi-year prediction of European summer drought conditions for the agricultural sector

“…It is the most commonly used skill measure to assess the contribution of the internally generated climate variability component to the multiannual predictive skill (Siegert et al 2017). However, it has been pointed out by some studies (Siegert et al 2017, Smith et al 2019 that using correlation difference for two highly correlated hindcasts underestimates the impact of initialization. Therefore, in order to determine whether initialization adds any additional information compared to the uninitialized forecast in the areas where both INIT and NoINIT return high correlation values, we apply the residual correlation methodology recently suggested by Smith et al (2019).…”

“…These decadal predictions are initialized with observation-based data and then run for a decade under the influence of contemporaneous changing external forcings, similarly to climate projections. The evolution of the climate system in this case is impacted by both internally generated and slowly varying natural variability and externally forced components (Meehl et al 2009, Doblas-Reyes et al 2013a, Meehl et al 2014, Smith et al 2019. A clear benefit of this new source of information is the potential to provide reliable and robust climate outlooks on a multi-annual timescale compared to long-term climate projections or climatological forecasts , Hewitt et al 2017.…”

Decadal Predictions of the Probability of Occurrence for Warm Summer Temperature Extremes

“…Recently, studies have shown that several climatic parameters are predictable up to 10 years into the future (e.g., Boer et al, 2016;Marotzke et al, 2016;Yeager & Robson, 2017). High-impact events like tropical (Dunstone et al, 2011) and extratropical (Schuster et al, 2019) storms, Sahel summer rainfall (Sheen et al, 2017), or Eurasian precipitation and sea level pressure (SLP) (Smith et al, 2019) were shown to be predictable on this time scale. Seasonal predictions of temperature extremes were shown to be skillful (e.g., Eade et al, 2012), as well as decadal predictions of warm summer temperature extremes-skill in the latter, however, was attributed to the global warming trend in the analyzed data (Hanlon et al, 2013).…”