Signal and noise in regime systems: A hypothesis on the predictability of the North Atlantic Oscillation

Abstract

Studies conducted by the UK Met Office reported significant skill in predicting the winter North Atlantic Oscillation (NAO) index with their seasonal prediction system. At the same time, a very low signal-to-noise ratio was observed, as measured using the "ratio of predictable components" (RPC) metric. We analyse both the skill and signal-to-noise ratio using a new statistical toy model, which assumes NAO predictability is driven by regime dynamics. It is shown that if the system is approximately bimodal in nature, with the model consistently underestimating the level of regime persistence each season, then both the high skill and high RPC value of the Met Office hindcasts can easily be reproduced. Underestimation of regime persistence could be attributable to any number of sources of model error, including imperfect regime structure or errors in the propagation of teleconnections. In particular, a high RPC value for a seasonal mean prediction may be expected even if the model's internal level of noise is realistic.

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

“…Analysis of the monthly NAO index based on the observations and 40 CMIP5 historical simulations further supports previously published results associated with the existence of the signal-to-noise paradox in the North Atlantic region (e.g., Dunstone et al, 2016;Eade et al, 2014;Scaife et al, 2014;Siegert et al, 2016;Strommen & Palmer, 2019). Over 75% of CMIP5 models have RSC values greater than 1.0, implying widespread existence of the signal-to-noise paradox in coupled climate models.…”

“…Recent advances in the development of seasonal climate prediction system have suggested a potentially robust level of forecast skills of the NAO variations, with the state-of-the-art initialized coupled climate model hindcasts and the realizations of large ensemble members (Athanasiadis et al, 2017;Baker et al, 2018;Riddle et al, 2013;Scaife et al, 2014;Stockdale et al, 2015). In the meanwhile, a remarkably lower signal-to-noise ratio in models compared with a high observed correlation score or a signal-tonoise paradox has been first pointed out by Scaife et al (2014) and subsequent studies (e.g., Dunstone et al, 2016;Eade et al, 2014;Siegert et al, 2016;Strommen & Palmer, 2019; and among others). Here we apply the statistical Markov model introduced in section 2 to determine if the signal-to-noise paradox occurs in the monthly NAO index simulated by 40 CMIP5 models and if the existence of the paradox is dependent on model initialization processes or just a general model problem.…”

Jet latitude regimes and the predictability of the North Atlantic Oscillation

“…This implies that one possible explanation is one or more overly weak teleconnections in the model. Both North Atlantic SSTs (Rodwell et al 1999;Mehta et al 2000;Gastineau et al 2013;Scaife et al 2013) An alternative hypothesis was proposed in Strommen and Palmer (2019), based on the idea that the NAO is driven by regime dynamics. The possible influence of regimes on North Atlantic circulation has been studied since the 80's, often by way of classifying distinct, frequently recurring and quasi-persistent geopotential height patterns, of which the two phases of the NAO are examples (Vautard and Vautard 1990;Michelangeli et al 1995;Dawson et al 2012).…”

“…Studies have shown that both frequency of occurrence of individual regimes and the transition rates between different regimes can be influenced by external forcing (Charlton-Perez et al 2018); it was suggested in Palmer (1999) that anthropogenic forcing may also influence regime behaviour in a similar manner. In Strommen and Palmer (2019), a statistical model was presented which captures the predictability inherent to such regime dynamics on seasonal time-scales. By envisaging the atmosphere as a two-state system, with transitions taking place on daily time-scales, it was shown that interannual variability in such a system is determined by changes in the preferred regime state each year and that this preference is in turn driven by variations in the persistence time-scales of the two regimes.…”

“…Given the fundamentally trimodal structure of the jet latitude, we would argue that a more appropriate way to analyse the contribution from jet latitude variations is through regime dynamics. As explained above, in the framework of the regime model from Strommen and Palmer (2019), the predictable component is not the average seasonal jet latitude itself, but rather the average seasonal persistence time-scales of the jet latitude regimes (which in turn determine the preferred jet latitude regime). By truncating the trimodal structure to a two-state system, we show that the IFS has skill at predicting seasonal shifts in these persistence time-scales, and hence that the signal-to-noise paradox can be explained by NWP models having systematic biases in regime persistence, as proposed in ibid.…”