Abstract ID: 233
Representation and Predictability of Stratospheric Wave Reflection Events in Subseasonal Forecast Models
Lead Author: Jason Furtado
School of Meteorology, University of Oklahoma, United States of America
Keywords: Stratosphere-troposphere coupling, Wave reflection, Subseasonal prediction, Extreme winter weather
Abstract: Variability of the Arctic stratospheric polar vortex constitutes one important subseasonal predictor for winter weather across the mid- and high-latitudes of the Northern Hemisphere. While variability in the vortex can manifest in several ways, recent work by the authors and others has focused specifically on the stratospheric wave reflection mechanism – i.e., upward-propagating Rossby waves that enter the stratosphere and subsequently reflect back downward into the troposphere, which can directly alter the jet stream and potentially induce extreme weather patterns across North America. While such events evolve on synoptic timescales, the potential skill for long-lead predictions of favorable environments for these stratospheric wave reflection events remains unknown.
This study directly addresses this knowledge gap by assessing how well subseasonal-to-seasonal (S2S) operational prediction systems simulate the fundamental characteristics and evolution of stratospheric wave reflection events. We use a published database of stratospheric wave reflection events (Messori et al. 2022) and a wave reflection index as bases for our verification. Using reforecasts from several models within the S2S Prediction Project Model Database, we first examine fundamental statistics of simulated stratospheric wave reflection events– e.g., seasonal frequency, amplitude, and spatial patterns of wave propagation associated with them. Biases between the simulated event statistics and those derived from ERA5 reveal that the models may overestimate the frequency of these events while also underestimating their magnitude. Additionally, the tropospheric weather patterns following these simulated wave reflection events are less consistent with slightly different teleconnections in some models versus reanalysis. Then, using about 20 observed events which overlap with the reforecast time period, we calculate several forecast skill metrics as a function of lead time (Weeks 1 to 4) to illustrate how far in advance such events might have been predicted and with what accuracy. Forecast error maps of tropospheric weather variables associated with the lagged response to the wave reflection events (e.g., 2-m temperatures, geopotential heights) are also presented for assessment, as ultimately these variables are of importance for stakeholders and the public. A look at how these results from reforecasts can be used for improving real-time subseasonal forecasts of Northern Hemispheric wintertime weather regimes will also be discussed.
Co-authors:
Oliver T. Millin (School of Meteorology, University of Oklahoma)
