Identifying and Assessing Gaps in Subseasonal to Seasonal Prediction Skill

“Estimates of predictability together with calculations of current prediction skill are often used to define the gaps in our prediction capabilities on subseasonal to seasonal timescales and to inform the scientific issues that must be addressed to build the next forecast system. However, different methods for estimating predictability can produce substantially different estimates of the upper limit of skill, leading to different conclusions regarding the gaps in our prediction capabilities. This project proposes to (1) systematically quantify estimates of the upper limits of predictability, (2) assess similarities and differences between predictability estimates and understand the reasons for differences between them, and (3) compare predictability estimates with current skill to identify gaps in our prediction capabilities.

To accomplish these objectives, we will apply several methods for estimating predictability to the North American Multimodel Ensemble (NMME) data as well as several other publically available datasets and compare them with each other and current prediction skill globally for monthly, seasonal and week 2-4 forecasts and for measures of specific phenomena (e.g. El Niño and the Southern Oscillation, Madden-Julian Oscillation, North Atlantic Oscillation). Additionally, we will evaluate extreme temperature, precipitation, and associated circulation anomalies. The outcome will provide information on the regions, variables, and phenomena where predictability estimates agree about whether current skill has or has not reached the limits of predictability and provide information on when and where we do not know the predictability limits. In the situations where the predictability limit is not clear because of disagreement in the estimates, we will diagnose the signal and noise from the methods to understand the source of the discrepancy. Where there is clear potential for more skill that has yet to be realized, the variability of the signal and noise from forecast to forecast, and its relationship to prediction skill will be evaluated to better understand current gaps in prediction capabilities.

Relevance: This project focuses on North American Multi-Model Ensemble System evaluation and application by assessing predictability and identifying the gaps in current prediction capabilities using the NMME re-forecasts. The analysis will evaluate predictability and current prediction skill globally and for specific phenomena, including extreme events such as heat waves, cold spells, and extreme precipitation anomalies. The focus on weeks 2-4 is a current priority for NOAA, and the results of this work will have a bearing on plans for model-based forecasts to operational climate prediction.”