14C.4 Early Season Weak Stratospheric Vortex Events in S2S Forecasts: Hits, Misses, and False Alarms

Thursday, 16 January 2020: 4:15 PM
257AB (Boston Convention and Exhibition Center)
Andrea L. Lang, University at Albany, SUNY, Albany, NY

Variability in the high latitude stratospheric flow can result in an increased probability of cold air outbreaks and anomalous shifts in the tropospheric storm tracks on subseasonal time scales during the cool-season. This type of coupling between the troposphere and stratosphere is well documented to occur during sudden stratospheric warming events, which are characterized by the reversal of the stratospheric westerlies to easterly and typically occur in mid-winter (DJF), but can also occur with other types of stratospheric variability. Since the tropospheric impacts of stratospheric variability can linger for up to 60 days after the stratospheric event, assessing forecasts of stratospheric variability is known to be important part of assessing subseasonal-to-seasonal (S2S) forecasts in the cool-season.

This analysis considers Northern Hemisphere stratospheric variability in S2S forecasts that verify in November. November is earlier than the climatological peak of troposphere-stratosphere coupling, which peaks in January and February, but the early season stratospheric variability can be associated with anomalous winter conditions. The analysis focuses on early season weak vortex events, defined to occur in the month of November when the 3-day averaged 10-hPa zonal-mean zonal wind at 60˚N is ≤ –1 standard deviation (approximately 10 m s–1). The research uses ERA-Interim reanalysis data as verification of S2S ensemble reforecasts of early season weak vortex events, specifically to evaluate the rate of forecast hits, misses, false alarm, and correct null forecasts at lead times ranging from 0 to 30 days. The analysis considers ensemble reforecast data from five operational models contributing to the WWRP/WCRP S2S Prediction Project Database common reforecast period (1996-2010). The results show that the rate of forecast hits exceeds the false alarm rate to lead times of 21-25 days for a majority of models. The forecast hits, misses, and false alarms are discussed in the context of model top height, resolution, and model configuration.

- Indicates paper has been withdrawn from meeting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