Polar Vortex Forecasts

All charts are updated once daily between 7-11 AM London time. Please feel free to share any images from this site on social media or elsewhere, with credit where appropriate.

You may need to refresh your browser to see the updated images. Click to see a full-size version of each image.

00Z GFS 0.5° & GEFS 1.0° 10 hPa 60°N [U] ensemble plume 

Climatological values use the full ERA-Interim record (1979-2019). Members exceeding 41.2 m/s are classified as having a “strong vortex event” following Tripathi et al. 2015, whilst those with winds below 0 m/s indicate either a major sudden stratospheric warming (SSW) or the final stratospheric warming (FSW).

Please note that the GFS and GEFS are currently very different models, so it might be expected that they noticeably diverge at longer lead-times. The GFS was upgraded to the new FV3 dynamical core in 2019, whilst a similar update to the GEFS is expected in 2020.

GEFS & GFS ensemble

00Z GFS 0.5° 10 hPa 60°N [U] forecast evolution

White crosses indicate a strong vortex event (>41.2 m/s) following Tripathi et al. 2015. A good forecast would form an invariant vertical line (i.e. with decreasing lead-time, the forecast remained constant).

GFS forecast evolution

00Z GFS 0.5° 10 hPa wind and geopotential height forecast maps 

Click here (opens in a new tab) to view/animate forecast maps of the 10 hPa vortex.

00Z GFS 0.5° 40-80°N average departures from zonal-mean geopotential height 

Click here (opens in a new tab) to view/animate a vertical cross-section.

CFSv2 1.0° 10 hPa 60°N [U] bias-corrected ensemble plume

This is produced following the archiving style used in the S2S database, i.e. today’s 00Z initialisations are combined with the 06, 12, and 18Z initialisations from yesterday to create a time-lagged 16-member ensemble, out to 44 days. The bias is calculated as the difference between the 1999-2010 hindcasts and ERA-Interim reanalysis. The CFSv2 has a significant bias toward a weak vortex (shown here as a function of launch date), especially for forecasts launched in early winter, so accounting for this drift is necessary.

It should be noted that a simple linear bias correction does not necessarily solve the problems produced by the bias, as the bias can interfere with the model dynamics.

CFSv2 ensemble

00Z GFS 0.5° & GEFS 1.0° 1000 hPa Northern Annular Mode (NAM) ensemble plume

This is calculated using the method of Gerber and Martineau (2018) using standardized anomalies of 65-90°N geopotential height (with respect to 1979-2018 ERA-Interim climatology) with the global mean anomaly removed. I thank Zac Lawrence for calculating the filtered ERA-Interim climatology used here.

NAM ensemble

00Z GEFS 1.0° 100 hPa 60°N [U] tercile categories

The tercile-category anomaly of the lower-stratospheric polar vortex, defined by the 100 hPa 60°N zonal-mean zonal wind, has been used as a diagnostic for the behaviour in the ‘coupling layer’ between the stratosphere and troposphere, i.e. the level in the stratosphere where circulation anomalies are important for influencing tropospheric weather regimes (e.g. Charlton-Perez et al. 2018, Lee et al. 2019). The chart below shows the percent of GEFS members in each tercile anomaly category, based on daily 1979-2018 ERA-5 climatology.

Weak = lower tercile; neutral = middle tercile; strong = upper tercile.

U100-60 GEFS terciles

00Z GFS 0.5° & GEFS 1.0° Scandinavia-Greenland dipole ensemble plume

This dipole pattern, defined in Lee et al. (2019) as the MSLP difference between a grid box over Scandinavia and a grid box over north-east Greenland, can be used as a diagnostic of anticyclonic wave breaking in the north-east Atlantic which can enhance vertically-propagating wave activity and weaken the stratospheric polar vortex. In the paper, we use 40 hPa as a threshold (strong Scandinavia high + deep Greenland low) as this is similar to what occurred before the February 2018 major SSW, but values above 30 hPa are noteworthy.

S-G ensemble

About the data processing
The data are retrieved from the NOAA NOMADS server. Issues with NOMADS may cause problems with data updates. The data analysis is performed on the Reading Academic Computing Cluster (RACC) at the University of Reading, primarily using Python. Issues with the RACC or other university IT services may interfere with what is shown on this website. It is possible to check the status of the RACC.

If you experience any issues with the site, or have any suggestions, please contact me: s.h.lee@pgr.reading.ac.uk.