Two polar lows, which appeared off the coast of Norway in November 2015, could be identified using ASCAT winds and satellite imagery.
14 November 2020
18 November 2015
By Sjur Wergeland (Norwegian Meteorological Institute)
In Winter 2015 forecasters at the Norwegian Meteorological Institute (met.no) got a new model for forecasting in the Norwegian Sea and the Barents Sea called AROME-Arctic. Since the model was fairly new forecasters needed to gain experience of its behaviour in different weather situations.
On the morning of Thursday 19 November, a perfect test situation occurred when two polar lows (PL) appeared in the northern part of the Norwegian Sea.
Since a polar low weather situation is not straightforward to forecast Sjur Wergeland from met.no decided to take the opportunity to validate the text forecast for Friday 20 November, issued at noon on the 19th.
With the aid of ASCAT winds the text forecast in a polar low weather situation for the area 'Banks outside Troms' was validated.
For the same area the wind forecast from four common used models at met.no were also validated. The ASCAT winds were also used to validate the +41h model forecast for the mean sea level pressure from three models frequently used at met.no.
The two polar lows can be clearly seen off the Norwegian coast in the Metop-B and NOAA-18 satellite imagery (Figures 1 and 2). In the northerly cold air outbreak the cumulus convection with higher cloud tops can be seen to the south.
Figure 2 shows the two polar lows were situated in an elongated trough in the HIRLAM 12 km isobars. The prognostic +6h sounding from between the two PLs shows deep instability in that area, up to about 475 hPa. Usually for PLs to develop instability up to at least 500 hPa is needed. The red 50 in the same area is the difference in degrees Celsius between SST and the temperature in 500 hPa — a common measure for the degree of instability. A threshold value of 44 degrees is usually needed for PL development to occur. The green lines are 500 hPa isohypses. Using them we can see that the PL intensification took place ahead of the 500 hPa trough, which is the normal place for PL development. The isohypses also indicate that the PL would move towards the northeast with the gradient wind. Northeast of Iceland there is a nice example of an Arctic front.
There are usually very few ordinary observations (synops) from the banks off northern Norway, but Friday afternoon we had ASCAT winds in the actual area. These helped validate the text forecast and different numerical weather forecast models.
At the forecasting centre (MWO) in Tromso we seldom use satellite pictures from Meteosat because of our northerly location the image resolution is quite low.
However, the Night Microphysics image from 19 November 08:30 UTC (Figure 3) shows most of the same features as the NOAA image, but in a much lower resolution. The polar lows show up in dark red as cold and thick ice-clouds.
The weather situation
On Figure 4 the left map is an analysis from Thursday morning. The low seen just to the west of Tromso is one of the two polar lows.
The other, southern polar low is visible on the analysis as a trough (marked as Minor Polar Low on the analysis). This moved towards south and weakened. The remaining polar low intensified as it moved towards the northeast.
The weather map on the right in Figure 4 is a 30-hour prognosis for 12:00 UTC on 20 November. It shows the remaining polar low (on the prognosis marked L) just to the north of Tromso.
Polar lows typically develop and intensify in a northerly, a so-called cold air outbreak. In a cold air outbreak Arctic air from above the polar ice cap is transported southwards by northerly winds and is eventually situated above a relatively warmer sea surface. This transforms the initially very dry and stable Arctic air into very humid and unstable maritime air, where a polar low can develop and stay.
The cold air outbreak is showed on the weather maps. The northerly winds were initially set up by the polar front low seen in the Barents Sea, southeast of Spitsbergen, on the left-hand weather map.
The bottom image on Figure 4 is a marinogram, a graphical forecast. It shows the evolution of wind, waves and surface current for a site known as the 'Banks outside Troms', just off the coast of Tromso.
During Friday afternoon and evening the wind was forecast to increase to north-westerly gale force 8, as the site was situated to the west of the polar low.
We do not forecast the steepness (the ratio wave height divided by wave length) of the waves. From the marinogram it can be seen that early on Saturday the waves and the surface current moves in opposite direction, a situation that usually gives quite steep waves.
Figure 5 shows the actual sea surface temperature (SST) and the ice off the east coast of Greenland. On the inset image the upper left figure is the sea ice situation and the upper right shows the colour code of the different ice categories. The red lines are SST in degrees Celsius. The location of the forecast areas is also shown.
The SST of 6–8 °C produced large upward fluxes of both heat and humidity into the air of the cold air outbreak, because of the strong destabilisation of the airmass.
This polar low diagram summarises the polar low climatology for the sea areas outside northern Norway. Inset on the diagram the bar chart shows that the sea areas have the highest numbers of polar lows in March, the time of the year when the Arctic air is at its coldest.
Some polar lows will eventually reach the coast; most of them do that along the coastline marked in red.
Validation of the text forecasts for the 'Banks outside Troms'
In a polar low weather situation the numerical weather models usually have different forecast scenarios, so the forecasters do not have an easy task in making good wind forecasts. That was also the situation on the morning of 19 November.
The different models we usually use produced, broadly-speaking, two different scenarios for Friday afternoon and evening. Figure 6 shows the wind forecast for 20 November at 15:00 UTC, from those different numerical weather models.
From Figure 6 we can identify two forecast scenarios:
- EC, Hirlam8 and Hirlam12 forecast north westerly winds of Force 6 to 8.
- The two (non-hydrostatic) models AROME-Arctic and AROME-METCoOp forecast south-westerly winds of Force 6.
From personal communication with the forecasters on duty I learned that they were very uncertain regarding which model to use. In the end they went for the two non-hydrostatic AROME-models, because a polar low is a highly convective phenomenon with quite large vertical accelerations and so probably best described in non-hydrostatic models. Hydrostatic models like Hirlam8 and Hirlam12 do not allow vertical accelerations.
High Seas Forecast, Thursday 19 November
HIGH SEAS BULLETIN FOR METAREA 19
Valid from 11:00 UTC on Thursday 19 November 2015
BY TROMSOE METEO, NORWAY
Banks outside Troms
VRB 4 to 6, FRI OCNL gale 8 FM between SW and NW. Snow SHWRS with poor.
As can be seen the forecast that was issued (in the blue box above) was somewhat indefinite, a consequence of the uncertainty in the numerical weather prediction models (NWPs).
Which of the models performed best for Friday afternoon?
Figure 7 shows the ASCAT observations for the 'Banks outside Troms' for three different times on the actual date.
Since the difference between the NWPs were large only for Friday afternoon the best models were those which forecast northwest near gale force 7 at that time — Hirlam8 and Hirlam12 closely followed by EC. So on this occasion the non-hydrostatic models were not the best ones in this convective situation.
And how good was the text forecast?
By using information from the ASCAT winds a nearly perfect text forecast regarding the wind should be something like “South force 5, in the forenoon veering west force 5 to near gale 7, strongest wind in the south. In the afternoon northwest near gale 7.”.
For comparison the official forecast was: "VRB 4 to 6, FRI OCNL gale 8 FM between SW and NW", which contains these issues:
- The not so precise term 'VRB'.
- Forecast gale 8, one Beaufort measure more than the ASCAT winds.
- The strongest wind was forecast nine to 12 hours too early.
- Ideally, perhaps, the phrase 'between SW and NW' should not have been used. It would have been better to use northwest only.
That said I think the text forecast was quite good, when you consider the polar low weather situation.
Validation of the mean sea level pressure numerical forecast
From Figure 8 we can see two small anticlockwise circulations; both are associated with the two polar lows marked on the image. That information can be used to validate the numerical weather prediction models.
We usually only get a few ordinary observations from these areas, so without the ASCAT winds such validation is virtually impossible to carry out.
Figure 9 shows the 41 hours forecast mean sea level pressure from three models frequently used at met.no.
The comparison with reality (upper left) is not very impressive. All models forecast the primary polar low too far towards the south west.
All models forecast a secondary polar low, but the position of this was, as could be expected, not perfect.
Figure 9 shows that the best model to forecast the secondary polar low, both position and circulation strength considered, was Hirlam8km. So a hydrostatic model was better than the non-hydrostatic AROME in this highly convective weather situation.
Significant wave height forecast
Figure 10 shows the significant wave height forecast (+42h) from two different numerical models, an observation from a rig situated at the forecast area 'Tromsoeflaket', and the wave height text forecast.
At the rig-observation site both models forecast a significant wave height of 2.5 m and that is in good agreement with the observation of 2 m.
The text forecast had 2 to 3 m in the area of the observation site, a quite good agreement with the observation.
For Deep D4 the two models show a difference of 2 m in maximum wave height, they also show different locations of the maximum value.
The text forecast had 4 to 5 meter in Deep D4, in good accordance with the lower resolution model that was preferred that day.
Without observations it would be difficult to find out which of the wave models performed best.
On Figure 11 the left panel show the climatology of the surface currents off northern Norway. The currents predominately flow from the south, towards the north.
The strongest current is a part of the so-called 'Norwegian current'.
The right panel on Figure 11 shows the forecast surface currents — quite low values at the 'Banks outside Troms', much stronger northerly currents at 'Deep D4', at least in eastern parts.
The wave spectrum (Figure 12) shows two wave groups, marked A and B. Wave group A is the dominant one — wind generated waves (seas) moving towards south/south east with a period of about nine seconds and a significant wave height of 4.3 m.
The minor wave group, B, shows swell moving towards the north east with a period of about 15 seconds.
As can been seen from Figure 12 the seas were moving towards south/south east, and that is opposite to the direction of the typically surface current and this is also valid for the forecast current. So, the forecaster has to expect quite steep waves in the area.
Wind generated waves moving towards the south on the westward side of a polar low are quite common, and since the typical surface current come from almost the opposite direction, a forecaster must be prepared for very steep waves in that area.
The wave steepness is not forecast, but since it is an important wave feature, especially for small boats, the forecaster must be aware of this. The wave steepness is wave height divided by wave length and this ratio has no unit.
Due to the uncertainty of the polar low weather situation the text forecast was somewhat unprecise.
The strongest wind blew from NW while the forecast said 'from between SW and NW'.
In the text forecast the strongest wind, gale force 8, was forecast nine to twelve hours too early, because the forecaster did not choose the most correct weather prediction model.
Validation of the wind forecast showed that the best models were the hydrostatic Hirlam8 and Hirlam12. The forecaster wrongly chose a blend of the two non-hydrostatic models AROME-Arctic and AROME-METCoOp because they thought they would perform best in this highly convective weather situation.
Validation of the +41h mean sea level pressure forecast gave the result that all three validated models did forecast the position of the primary polar low poorly. The position of the secondary, and minor, polar low was best forecasted by Hirlam8.
So, both the timing of the strongest north-westerly winds behind the polar low and the position of the secondary polar low was best forecast by a hydrostatic weather prediction model.
Latest case studies
Devastating floods in western Europe
Catastrophic floods hit Germany, Belgium and parts of West Europe mid July 2021.
Catastrophic tornado in the Czech Republic
Satellites' view on the catastrophic tornado case on 24 June 2021 in the Czech Republic.
Tracking the Gulf Stream with satellite data
Using satellite data from multiple satellite instruments to track the Gulf Stream in 2020 and 2021.