The year 2020 was marked by high-quality weather predictions ECMWF provided to its Member and Co‑operating States and other users of its data and products across the globe. Severe weather events that were well predicted by ECMWF forecasts include wet and stormy weather in Europe in February, Hurricane Laura making landfall as a category 4 hurricane in Louisiana, USA, and a storm in October that affected large parts of western Europe, bringing significant wind gusts and high precipitation.

An upgrade of the Integrated Forecasting System (IFS) implemented in June improved global forecasts and substantially improved analyses and forecasts in the stratosphere. IFS Cycle 47r1 also introduced a new tropical cyclone wind radii product, which facilitated the identification of wind-related hazards. The year saw significant improvements related to upper-air ensemble forecast skill and surface parameters due to IFS Cycle 46r1, which was implemented in June 2019. Compared to other global modelling centres, ECMWF was able to maintain the overall lead in the medium range.

In an example of cooperation with the World Meteorological Organization (WMO), ECMWF joined up with the WMO to launch a new web-based interface in March 2020, to help monitor the availability and quality of global meteorological observations.

Windstorm Alex

At the beginning of October 2020, storm Alex affected large parts of western Europe, bringing wind gusts above 50 m/s to northwest France and precipitation locally more than 500 mm/24 hours to northern Italy and southeast France that resulted in multiple fatalities.

Storm AlexThis was the first storm of the 2020–21 European windstorm season, making landfall in Brittany, France, on 2 October 2020 while at peak intensity. (Image: EUMETSAT)

The most severe band of wind gusts over Brittany had some of the hallmarks of a sting jet. The maximum wind gusts in forecasts three days before (initialised at 12 UTC on 29 September) and onwards ranged from 44–51 m/s, compared to 51 m/s in observations. Such predictions give confidence that the model can simulate this type of fine-scale feature. In earlier forecasts, strong wind gusts were predicted, but not at the observed levels. Those gust forecasts had more broadscale characteristics, more reminiscent of a cold jet phenomenon (see the Extreme Forecast Index in the left-hand panel below).

Extreme Forecast IndexExtreme Forecast Index (EFI) for 24-hour maximum wind gusts on 1 October (left) and precipitation from 00 UTC on 2 October to 00 UTC on 5 October (right) in forecasts from 27 September. The plots also show areas as green lines where the values are greater than 0.9 in the one-day EFI maximum wind gust forecast on 1 October and the three-day EFI precipitation forecast starting on 2 October.

For the precipitation over southeast France and northern Italy, we find that the signal for this extreme rainfall event gradually became stronger, even if we see a slight jump on 29 September that coincides with the much clearer jump for the wind gust event (see the right-hand Extreme Forecast Index chart). The extreme rainfall was due to strong advection of moisture from the Mediterranean Sea towards the southern Alps, as seen in the specific humidity analysis.

Specific humidity analysisBackward air trajectories ending off the Mediterranean coast on 2 October 12 UTC between 925–850 hPa, based on ECMWF analyses and calculated with LAGRANTO, a software package to calculate parcel trajectories in the atmosphere provided to ECMWF by ETH Zurich. The colour of the trajectories indicates the specific humidity.

February’s wet and stormy weather

February 2020 was dominated by a strong westerly flow over northwest Europe. Several severe cyclones affected the region. These included storm Ciara on 8–9 February and storm Dennis on 15–16 February (as named by Met Éireann and the UK Met Office), and further systems on 22‑23 February and 28‑29 February. The cyclones brought strong winds and heavy precipitation leading to extreme accumulations over the month. New monthly precipitation records were set in England and Wales, Denmark and parts of southern Sweden, and there was some flooding in those countries.

The strong cyclonic activity over the North Atlantic was associated with a strong positive phase of the North-Atlantic Oscillation (NAO). An NAO+ signal for the week from 10 to 16 February was present in the extended-range forecast from 23 January and became stronger in the forecast from 30 January. The predicted flow-regime anomaly is visualised in a new two-dimensional plot of NAO-Blocking regime phases for the northeast Atlantic.

Although this product was originally designed to give early warnings about cold spells, it also highlights the likelihood of a westerly flow across the Atlantic with warm and wet weather over western Europe. The 30 January flow-regime anomaly forecast was shifted towards a combination of NAO+ and a trough over Scandinavia (Blocking–). The verifying analysis was in the same region of the diagram but somewhat more extreme.

Weather regime outlook for the northeast AtlanticThe plot shows the NAO-Blocking weather regime probability density function for an ensemble forecast starting on 30 January 2020 for the week from 10 to 16 February. Daily values of the verifying analysis are represented by dots: yellow (first day) to brown (last day of the target period).

Forecast performance

On 30 June 2020, ECMWF implemented Cycle 47r1 of the Integrated Forecasting System (IFS). This brought substantial improvements in the stratosphere as well as slight improvements in the troposphere.

ECMWF’s headline scores are computed as 12-month running averages to filter out the annual cycle and better identify trends in forecast performance. This means that the beneficial effect of new model cycles is fully visible only 12 months after implementation.

The first figure shows the significant improvement of upper-air ensemble forecast (ENS) skill due to IFS Cycle 46r1, which was implemented in June 2019. Results are shown for vector wind and temperature at 850 hPa and geopotential at 500 hPa.

The second figure shows that the beneficial effect of this cycle includes surface parameters, specifically a further reduction of the fraction of large ensemble forecast (ENS) errors in 2 m temperature. Substantial improvements were also seen in the precipitation forecast.

Compared to other global modelling centres, ECMWF was able to maintain the overall lead in the medium range, both for upper-air and surface parameters. It is worth noting that the medium-range forecast performance of the IFS did not show any obvious degradation due to reduced aircraft observations from March 2020 onwards as a result of COVID-19.

The signal was apparently sufficiently small to get masked by natural performance variations within the annual cycle, year-to-year atmospheric variability, as well as the positive effects of new and additional observations and the most recent model upgrade.

Upper-air ENS skill improvementsSkill of the ENS at day 5 for three upper-air parameters in the northern extratropics, relative to a Gaussian-dressed ERA5 forecast as a reference. Values are running 12-month averages, and verification is performed against own analysis.
Reduction in the occurrence of large ENS 2-metre temperature errorsEvolution of the fraction of large 2-metre temperature errors (CRPS > 5K) in the ENS at day 5 in the extratropics. Verification is against SYNOP observations. The 12-month running mean is shown in red, the 3-month running mean in blue.

Wind radius of tropical cyclones

To assess wind-related hazards associated with tropical cyclones, it is useful to see the areas where winds are predicted to exceed certain thresholds. In Cycle 47r1 of the Integrated Forecasting System (IFS), introduced in June 2020, this was made possible by introducing a new wind radii parameter. The parameter indicates the maximum distance from a tropical cyclone centre within which the surface wind speed is predicted to exceed certain thresholds.

The thresholds have been set at 34, 50 and 64 knots, in line with the values used by tropical cyclone warning centres. A module from the Vortex Tracker package developed at the Geophysics Fluid Dynamics Laboratory (GFDL) is used to compute the wind radii.

The wind radii computation is performed after the ECMWF tropical cyclone tracker has completed the identification of cyclonic features for both high-resolution forecasts (HRES) and ensemble forecasts (ENS). To start, the algorithm establishes four sectors (NE, SE, SW and NW quadrants) centred on each tropical cyclone’s predicted positions. The wind radii represent, for each sector, the maximum extent from the storm centre at which the wind thresholds are exceeded.

The figure shows an example of a high-resolution wind radii forecast for Hurricane Dorian starting from 12 UTC on 30 August 2019. It shows that 34‑knot winds are predicted within the sectors shown. Similar charts can be produced for 50- and 64‑knot wind radii if such wind speeds are present in the forecast.

Wind radii forecastHRES wind radii forecast (IFS Cycle 47r1) applied to Hurricane Dorian in 2019 as an example for the change made in 2020. The forecast shows the 34-knot wind threshold up to 240 hours ahead, initialised at 12 UTC on 30 August 2019. The red dots indicate the predicted centre of the hurricane at 12-hour intervals.

New web tool for monitoring the quality of observations

The World Meteorological Organization (WMO) and ECMWF launched a new web-based interface in March 2020 to help monitor the availability and quality of global meteorological observations. The new tool monitors the performance of the in-situ observing systems that are a key component of the WMO Integrated Global Observing System (WIGOS) and is part of the WIGOS Data Quality Monitoring System (WDQMS).

The WDQMS web tool monitors the availability and quality of land-based surface and upper-air observations, based on near-real-time monitoring information provided by four participating global numerical weather prediction (NWP) centres: the German national meteorological service (DWD), ECMWF, the Japan Meteorological Agency (JMA) and the US National Centers for Environmental Prediction (NCEP).

The new system is able to collect six-hourly quality monitoring reports from the four WIGOS Monitoring Centres and store the data in the WDQMS database at ECMWF. The data are then aggregated, and the calculated statistics are compared against performance thresholds. An important function of the tool is to routinely compare the number of observations delivered from the national meteorological services (NMS) all over the world with the number of observations the NMS were expected to deliver. Under this function, the system flags up any discrepancies between what was scheduled and what was observed and highlights any issues.

The WDQMS web toolThis screenshot shows the daily status of the land surface network for 17 March 2020 based on the monitoring information aggregated across the four NWP centres. The map highlights inhomogeneities in global data coverage (black means not reporting during the period) and reporting practices (orange and red mean underperforming and green means fully reporting) as well as issues in the station metadata (pink means reporting more than expected and yellow means that the station ID is not known).

Compared to other global modelling centres, ECMWF was able to maintain the overall lead in the medium range.