Polar Vortex displacements and mid-latitude responses

Based on a presentation given by Bethany to the Royal Meteorological Society’s Young Career Scientist Conference, July 2017.

What is the Polar Vortex?

The northern hemispheric polar vortex is a winter phenomenon.  It is an upper level region of low pressure and cold air lying near the earth’s North Pole which exists during the winter months (Figure 1) .

Figure 1: Geopotential height at 10 hPa over the North Pole in winter (left) and summer (right).  Source: NOAA ESRL PSD.

Weak Polar Vortex events and implications for mid-latitude temperatures

During events when the polar vortex weakens, it can either be displaced or split. Perturbations from weak vortex events can descend from the stratosphere down into the troposphere and influence surface temperature in the mid-latitudes but there is little obvious predictability between the type of weak vortex event and the location or severity of the temperature anomaly.

For example, a polar vortex displacement event in November 2016 gave negative surface temperature anomalies over Europe and northern Russia and positive surface temperature anomalies over North America (Figure 2).

Figure 2: Geopotential height anomaly at 10 hPa (left) and air temperature anomaly at 850 hPa (right) for 1st to 21st November 2016. Source: NOAA ESRL PSD.

Whereas, another polar vortex displacement event in March 1990 gave an opposite signal; with warm temperature anomalies over Europe and northern Russia and cold temperature anomalies over North America (Figure 3).

Figure 3: Geopotential height anomaly at 10 hPa (left) and air temperature anomaly at 850 hPa (right) for 10th February to 3rd March 1990. Source: NOAA ESRL PSD.

Classifying weak Polar Vortex events

Numerous methods have been used in the literature to classify different polar vortex events, often with somewhat arbitrary definitions of regions and/or thresholds.  In this study, Principal Component Analysis (PCA) is used. PCA identifies a small number of dominant patterns from a large data set which explain the maximum amount of variance with the fewest number of patterns. The first principal component accounts for as much variability as possible (e.g PC1 in Figure 4) and each succeeding component has the highest variance possible providing it is orthogonal to the sum of the preceding components (e.g. PC2 in Figure 4 ).

Figure 4: Principal Component Analysis (PCA) on a distribution with the longer arrow PC1 and the shorter arrow PC2. Source: Nicoguaro, 2016.

In this study, PCA was conducted using ERA interim reanalysis data from the last 35 years for geopotential height at the 10 hPa level over winter (here taken as November through March)  to find the main pressure patterns over the North Pole.

Five main patterns were identified from the analysis (Figure 5) and the results can be of both positive and negative amplitude.

Fig 5: Principal Components (PCs): top row, displacement events and bottom row, split events.  Source: LSC, charted using software from GISS NOAA

Principal component (PC) 1 is the winter vortex, as shown in Figure 1, PC 2 and PC3 are representative of displacement events and PC 4 and 5 are representative of split events. Table 1 shows the variance each PC accounts for.

Table 1: Percentage variance of each Principal Component (PC).

Having defined the main winter Polar Vortex pattern and four types of weak Polar Vortex patterns (two displacements and two splits), we classify each date within the 35 years into one of these 5 patterns, according to which it is most similar to.  When a displaced or split pattern was consistently dominant for five consecutive days or more, it was classed as an ‘event’.  From the years studied 145 displacement events and 28 split events were recorded.

Looking into mid-latitude temperature responses

To identify differences in mid-latitude responses to these weak Polar Vortex events, six regions (five landmasses and one ocean) were categorised between the latitude 40 and 70N (Figure 6).

Figure 6: Regions to identify mid-latitude responses. Charted using software from GISS NOAA

For each split or displacement event the average temperature anomaly at 850 hPa for each region was calculated.

Frequency, by region, of coldest mid-latitude response

Figure 7 shows the frequency with which the regions saw the coldest temperature anomaly at 850 hPa. (Note that displacement 1 is PC 2, displacement 2 is PC 3, and split 1 is PC 4 and split 2 is PC 5 as shown in Figure 5.)

Figure 7: The region with the coldest temperature anomaly at 850 hPa for each displacement and split event.

Displacement events appear to give a wider distribution of cold mid-latitude temperatures than split events.  Siberia, Alaska and North America appear to be more likely to see cold than Europe or Russia.

Are there correlations between strength of PV weak event and mid-latitude response?

Choosing a type 1 split event over Siberia as an example, is there a correlation between the strength of the event (measured by the amplitude of the PC) and the associated temperature anomaly?  From figure 8, we can see that there is evidence of a weak correlation.

The event circled in Figure 8 is an event of negative amplitude, with negative temperature anomalies over Siberia.

Figure 8: Correlation between the amplitude of a type 1 split event and the temperature anomaly over Siberia (left) and the positive amplitude of PC 4 (right).

The event circled (Figure 8) refers to an event in March 1988.

Figure 9: Geopotential height anomaly at 10hPa (left) and air temperature anomaly at 850 hPa (right) for 3rd to 9th March 1988. Source: NOAA ESRL PSD.

As expected, Figure 9 (left) shows the pressure pattern of PC 4 with a negative amplitude (reverse pressure of Figure 8 right) and a cold temperature anomaly over Siberia.

Concluding remarks

PCA clearly identifies northern hemispheric polar vortex displacement and split events.  Some regions, i.e. Siberia, Alaska and North America, are more prone than others to tropospheric cold air outbreaks in the mid-latitudes.

Next steps include looking at predictability: to what extent the results found in this study will aid the winter forecasting of mid-latitude cold outbreaks?