January 18, 2022 11:10 pm

Study of the storm Filomena and its extreme snowfalls

Francisco Martin Leon 11 min
Satellite image after the Filomena pass where large snowy areas are observed in the Iberian Peninsula on January 11, 2021 according to the Terra satellite. POT

RAM note. The intense and extensive snowfall that it generated the storm Filomena They were historical and extraordinary, reaching record thicknesses in many areas of the Peninsula. Its impact was extraordinary. After its passage a very intense cold wave affecting large areas of the peninsula, exacerbating and prolonging its effects.

The Madrid area recorded snow accumulations of up to 50 cm in 30 hours, which meant heaviest snowfall since 1971. After the disappearance of Filomena, an extreme cold wave affected the entire Peninsula. This can also be considered historical due to minimum temperature records in many areas.

The snow episode forced the closure of the Madrid-Barajas airport and caused the cancellation of all rail service in the region. Schools were also closed for several days.

The Spanish Meteorological Agency (AEMET) named the storm Filomena on January 5, when its center was near the Azores Islands. On January 8, Filomena moved in a northeast direction towards the south of the Iberian Peninsula as an extratropical storm structure (cold, warm and occluded fronts). Meanwhile, a very cold polar air mass was established with minimums below 0 ° C in practically the entire Iberian Peninsula.

Surface analysis. Excerpt from the AEMET surface analysis at 12 UTC January 8, 2021.

Therefore, when the storm Filomena reached the Peninsula, the warm and humid air moved over the very cold air below (see the AEMET graph). With the exception of some southern coastal areas, all rainfall occurred as snow during January 8 and 9, which resulted in a layer of snow throughout the central and northeast quadrant of the Peninsula after the storm. Red warnings were set in most of those areas due to the 24-hour high rainfall accumulations expected in the ECMWF Ensemble Forecast (ENS) for several consecutive model runs.

Snowfall accumulations

The high resolution forecast from the ECMWF (HRES) and the ENS over Madrid provided some indication of the risk of snowfall from the beginning of January.

From January 3 at 12 UTC, the median ENS indicated a total snowfall of 20 to 30 mm for the period Jan 7-9, compared to around 10mm in previous forecasts. Even in previous forecasts, a substantial number of ENS members indicated large amounts of snow. However, a large dispersion was observed in the middle range due to the uncertainty regarding the exact location of the convergence between the polar and subtropical air masses.

The Extreme Forecast Index (EFI) was variable according to the changes in the forecast of the whole; However, a clear sign of a A rare snowfall event was present six days in advance with EFI values ​​between 0.7-0. 8 in Madrid (see EFI figure and box-and-whiskers forecast, box-and-whiskers, for the area around Madrid).

RAM note. Box and Whisker Plot

Box-whisker diagrams (boxplots or box and whiskers) are a visual presentation that describes several important characteristics at the same time, such as dispersion and symmetry.

For its realization, the three quartiles and the minimum and maximum values ​​of the data are represented, on a rectangle, aligned horizontally or vertically.

EFI values ​​were close to 1 four days ahead, with some uncertainty in the exact spatial distribution of the anomalous climate. The 72-hour average observed precipitation to January 10, 00 UTC between 11 stations within a 1×1 degree box ranged from 5 to 53 mm, showing great variability in the distribution of snowfall. This compares with more than 60 mm forecast in the south of the box and between 10 and 20 mm in the north by the corresponding HRES forecast as of January 5 00 UTC (see the observations figure and the HRES forecast).

The same January 6 forecast showed generally higher amounts of precipitation in southeastern Spain. However, a considerable decrease in the amount of precipitation predicted was again observed in the shorter lead time in the run on January 7, 00 UTC, mainly due to the uncertainty in the cyclone path between consecutive model runs. It should be noted that there is also great uncertainty in the observations, since the SYNOP stations in Spain cannot correctly measure snowfall due to the lack of heating of the rain gauges. Therefore, we expect the observed precipitation to be underestimated, making this analysis only an estimate.

Extreme Forecast Index (EFI) and rainfall forecast for the Community of Madrid. The EFI is shown with a start time of 00 UTC on Jan 3, 2021 for total snowfall on Jan 7-9, 2021 (left). The box-and-whisker plot (right) shows the evolution of precipitation forecasts in the same period in the box marked around Madrid in the left panel for different starting dates.

Very cold conditions after Filomena

When the North Atlantic Oscillation (NAO, abbreviated in English) enters a negative phase usually has a great impact on the Iberian Peninsula. In this phase, the high-level jet stream moves to lower latitudes, resulting in above-normal rainfall and lower temperatures in southern Europe.

From the end of 2020, global models began to forecast a possible negative NAO in early 2021, and this situation may have contributed to creating wet and cold conditions in almost the entire Iberian Peninsula during Filomena.

After Filomena, many areas in northeastern Spain fell to record low temperatures for several days in a row. Temperatures from -20 to –26 ° C were recorded in some areas with anticyclonic conditions. The figure showing the forecast for temperature anomalies in Europe up to two weeks earlier indicates that the low temperatures were part of a larger phenomenon. Madrid registered a low of –6 ° C on January 12. The 72-hour average temperature forecast through January 10 00 UTC within a 1×1 degree box suggests a downward trend in temperature for those days from early January.

In the forecasts for January 9-10, the median of the ENS indicated average temperature values ​​within this box close to –5 ° C. However, the forecasts issued on January 10 for the same day increased these values ​​to around –2 ° C and reduced the uncertainty of the forecasts. We can also systematically observe higher temperatures in the HRES than in the ENS Control for different waiting times of the event (see the temperature forecast for the Madrid region). An evaluation of the model’s performance shows that temperatures were generally overestimated in many areas of the interior of the Peninsula.

Observed and predicted cumulative precipitation. The graphs show the accumulated precipitation from January 7, 00 UTC to January 10, 00 UTC in the observations (left) and in the HRES forecast from January 5, 00 UTC (right).
Forecast of temperature anomalies two weeks in advance. The graph shows the 2m temperature anomaly forecast in Europe initialized at 00 UTC on January 4, 2021 and valid from January 11-17, 2021.

Overall, the IFS successfully predicted this historic snowfall event well in advance. The forecasts included the normal uncertainties associated with the exact location of the convergence of the polar and subtropical air mass and the total number and extent of the worst affected areas, but contained enough information to provide relevant weather warnings well in advance.

Temperature forecast for the Madrid region. The box-and-whisker plot shows the evolution of the 2 m temperature forecasts for the 72-hour period from January 11 to 13, 2021 in the box marked in the figures above for different start dates.

Estíbaliz Gascón Linus Magnusson (both ECMWF) Jesús A. Barroso (AEMET, Spain)


This entry was published in Reports on Nov 26, 2021 by Francisco Martín León


Comments (0)

Leave a Reply

Your email address will not be published. Required fields are marked *