How confident are predictability estimates of the winter North Atlantic Oscillation?

Abstract

Atmospheric seasonal predictability in winter over the Euro-Atlantic region is studied with an emphasis on the signal-to-noise paradox of the North Atlantic Oscillation. Seasonal hindcasts of the ECMWF model for the recent period show, in agreement with other studies, that correlation skill over Greenland and parts of the Arctic is higher than the signal-to-noise ratio implies. This leads to the paradoxical situation where the real world appears more predictable than the models suggest, with the forecast ensembles being overly dispersive (or underconfident). However, it is demonstrated that these conclusions are not supported by the diagnosed relationship between ensemble mean root-mean-square error (RMSE) and ensemble spread which indicates a slight under-dispersion (overconfidence). Furthermore, long atmospheric seasonal hindcasts suggest that over the 110-year period from 1900 to 2009 the ensemble system is well calibrated (neither overnor under-dispersive). The observed skill changed drastically in the middle of the twentieth century and paradoxical regions during more recent hindcast periods were strongly under-dispersive during mid-century decades. Due to non-stationarities of the climate system in the form of decadal variability, relatively short hindcasts are not sufficiently representative of longer-term behaviour. In addition, small hindcast sample size can lead to skill estimates, in particular of correlation measures, that are not robust. It is shown that the relative uncertainty due to small hindcast sample size is often larger for correlation-based than for RMSE-based diagnostics. Correlation-based measures like the RPC are shown to be highly sensitive to the strength of the predictable signal, implying that disentangling of physical deficiencies in the models on the one hand, and the effects of sampling uncertainty on the other hand, is difficult. Given the current lack of a causal physical mechanism to unravel the puzzle, our hypotheses of non-stationarity and sampling uncertainty provide simple yet plausible explanations for the paradox.

Subseasonal‐to‐Seasonal Predictability of the Southern Hemisphere Eddy‐Driven Jet During Austral Spring and Early Summer

“…More specifically, it allows for a continuous rather than a discrete representation of the signal. A potential disadvantage of this alternative signal‐to‐noise model is that recent work has suggested that some forecast models used in numerical weather prediction may be overdispersive (Scaife & Smith, ; but see also Weisheimer et al, ; Osman et al, ). If such an overdispersive scenario was the case for the present hindcast ensemble, then this alternative signal‐to‐noise model would offer an unduly pessimistic estimate of S2S forecast skill.…”

When and where do ECMWF seasonal forecast systems exhibit anomalously low signal‐to‐noise ratio?

“…(), during recent winters there has been a large increase in the variability of the NAO and analysis of hindcasts initialized over a 110‐year period (Weisheimer et al . ) suggests that there is significant variation of over‐ and underconfidence during the 20th century linked to long‐term variability of the global circulation and teleconnections (e.g. O'Reilly, ).…”

“…Weisheimer et al . () also see differences in over‐ and underconfidence when they are measured with different diagnostics, and suggest that, when the predictable signal in the hindcasts is too weak, some diagnostics of over‐ and underconfidence are more sensitive to a change in the amplitude of the signal than others. It is important to note, though, that the results of Weisheimer et al .…”

“…However, Weisheimer et al . () show that, at least over Greenland and calculated over a longer period, RPC is significantly greater than one in both System 4 and System 5. Weisheimer et al .…”

“…Weisheimer et al . () also show the large sampling variability of the RPC diagnostic when applied to short hindcast sets. For the AO, Stockdale et al .…”