J19.2 Ensemble Prediction and Predictability of Extreme Weather on Sub-seasonal to Seasonal Time Scales using Circulation Regimes (Invited Presentation)

Tuesday, 14 January 2020: 11:00 AM
104C (Boston Convention and Exhibition Center)
David M. Straus, George Mason Univ., Fairfax, VA; and K. Pegion

We seek to advance the predictive capability of extreme storminess and precipitation on the subseasonal to seasonal time scales, using multi-model ensembles of reforecasts and forecasts from the SubX multi-model subseasonal prediction experiment, the North American Multi-Model Ensemble, and NOAA’s Unified Forecast System ensemble forecasts. The approach is based on the ability of forecast models to realize the enhanced predictability of the planetary waves at subseasonal to seasonal time scales compared to the more limited predictability of synoptic scales, and the highly uncertain predictions of precipitation.

We identify preferred large-scale circulation patterns (circulation regimes) from reanalysis. We then evaluate “extremes metrics” based on the pdfs of precipitation, surface temperature and storminess (Eulerian storm track measures) for each large-scale circulation regime from observations. These metrics include anomalies, normalized anomalies, and ratio of times for which precipitation, temperature and storm-track strength are in the top (or bottom) 5thpercentile compared to the number of times expected solely on the basis of regime frequency. Evaluation of forecasts and reforecasts in terms of the observed circulation regimes then forms the basis of forecasts of extreme weather on sub-seasonal to seasonal time scales.

We review the organization of planetary waves by multiple, regional circulation regimes over both the Euro-Atlantic and Pacific-North American regions for overlapping three-month seasons from ERA-Interim reanalyses. For each regime and season, we evaluate the regime-dependent metrics of extreme weather established in detail by Amini and Straus (2019).

To categorize multi-model ensemble subseasonal forecasts in terms of regimes, bias corrected anomalies are formed for each model, after which pattern correlation is used to assign each ensemble member to a particular regime. The entire multi-model forecast is given in terms of the probability of occurrence of each regime, allowing us to flag a priori forecasts for which the potential predictability is high (good agreement among ensemble members for long-lasting regimes). The forecast of each extreme metric is based on the set of observed metrics for each regime, weighted by its probability of that regime occurring. For seasonal forecasts, the ensemble-predicted probability of occurrence of each regime is used to produce forecasts of weather extremes for the season. A comparison between these forecasts of extreme metrics with the forecasts obtained directly from the model output of precipitation and storminess diagnostics will test our method.

Amini, S., and D.M. Straus, 2019: Control of Storminess over the Pacific and North America by Circulation Regimes. Climate Dynamics,52, pp 4749–4770.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner