An Extratropical Contribution to the Signal‐To‐Noise Paradox in Seasonal Climate Prediction. Issue 23 (10th December 2022)

• Record Type: Journal Article
• Authors: Knight, Jeff R.; Scaife, Adam A.; Maidens, Anna
• Is Part Of: Geophysical research letters. Volume 49:Issue 23(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-10
• Subjects: signal‐to‐noise paradox -- seasonal prediction -- tropical‐extratropical teleconnections -- relaxation experiments; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2022GL100471
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: Skillful seasonal prediction of northern hemisphere mid‐latitude winter climate has become a reality in the last decade, but its practical applications are hampered by the "signal‐to‐noise paradox, " in which predicted signal amplitudes are much smaller than expected from the correlation with observations. Various hypotheses for the paradox have been advanced. Here we test whether we can identify a contribution from extratropical processes. To do this, we use ensembles of numerical experiments in which the tropical dynamical state is relaxed toward observationally based reanalysis. This allows us to remove errors in the representation of the tropical sources of teleconnections to mid‐ and high‐latitudes. We find that a signal‐to‐noise paradox remains present in our relaxation experiments, implying that at least part of the origin of the paradox must arise from errors in simulating processes in the extratropics. This finding helps to narrow the search for its ultimate cause. Plain Language Summary: Many economic sectors could benefit from skillful indications of mid‐latitude weather in the season ahead. For winter, at least, moderate skill has become a reality in recent years for some mid‐latitude regions. Despite this, these predictions suffer from a problem that limits their usefulness. Winter seasonal forecasts are reasonably successful in predicting the direction of the change in likelihood, for example, showing more chance of a cold winter when cold winters actually occur. However, the sizes of these shifts are much smaller than would be expected from this level of skill. This points to shortcomings in the climate models used to make the predictions. In this study, we test whether this is a result of errors in model processes occurring outside of the tropics. To do this, we artificially impose observed conditions in the tropical atmosphere of a seasonal prediction model to eliminate tropical errors that can influence mid‐latitude predictions. We find that the problem remains despite this intervention, implying its origin must, at least in part, be from errors originating in middle or high latitudes. Key Points: Tropical relaxation is used to test whether the mid‐latitude "signal‐to‐noise paradox" can be removed by specifying the tropical state This is found not to be the case, implying that at least part of the paradox is due to errors in simulating processes in the extratropics The results have implications for narrowing the identification of the source of the paradox