The Extreme Forecast Index (EFI) for water vapour flux became operational in Integrated Forecasting System (IFS) Cycle 46r1 in June 2019 (ECMWF Newsletter 160). The benefit of this EFI parameter is twofold: (1) potential to provide earlier awareness of extreme precipitation on the west coasts of mid-latitude continents than by using the EFI for precipitation; and (2) illustration of the synoptic-scale processes (atmospheric river activity) behind extreme hydrometeorological events. This article details the collaboration between ECMWF and the Flood Forecasting Centre (FFC) at the UK Met Office on this EFI parameter and its operational trial during the winter of 2019/20, and it introduces some useful guidance for its use and interpretation alongside other products.
Daily tasks
ECMWF scientists and FFC hydrometeorologists have discussed the forecast opportunities surrounding the EFI for water vapour flux in multiple meetings since 2016. Of most interest to the FFC is the possibility to have enhanced predictability of extreme events so they can give Category 1 and 2 Responders (such as the Environment Agency and Natural Resources Wales) additional time to prepare and respond to such events. Following a pre-operational assessment in winter 2018/19, an operational trial was conducted in winter 2019/20. This coincided with a protracted period of unsettled weather and widespread flooding. Throughout the winter, a daily task was scheduled to evaluate the water vapour flux EFI fields and to monitor regions near the UK where the EFI exceeded 0.5 (this value signifies the possibility for extreme conditions). The focus was predominantly on forecast days 6 to 10, and the EFI was used by FFC hydrometeorologists as an additional decision-making tool when writing the Flood Guidance Statement.
An example of the useful guidance provided by the EFI for water vapour flux is given in the figure for Storm Dennis on 15 February 2020. A week after Storm Ciara, Dennis brought further widespread heavy and persistent rainfall to high ground across much of England and Wales, with the largest rainfall accumulations over the Brecon Beacons in south Wales. The water vapour flux EFI gave useful guidance at days 7–9, and particularly from day 7, for another large-scale rainfall event that, as it would be falling on mostly saturated catchments, had potential for significant flooding impacts. This was an earlier and stronger sign than that provided by the precipitation EFI.
Lessons learnt
Lessons learnt from the trial include a number of useful guidelines for use of the EFI for water vapour flux from an operational perspective. It is first most appropriate to assess the EFI for water vapour flux and precipitation together to give FFC hydrometeorologists the best opportunity to understand the potential for rainfall/flood events. Second, during winter 2019/20, earlier indications of wet periods of weather (at day 6 and beyond) were found by using the EFI for water vapour flux compared to the precipitation EFI; moderate EFI values for water vapour flux were also occasionally seen on days 10–15. However, the EFI for water vapour flux was often rather too broad and, on occasions, shifted to the south on forecast days 6–9 compared to observed rainfall (and sometimes compared to the precipitation EFI). This southward displacement of the EFI signal was associated with a series of travelling lows. Third, the archetypal synoptic model for atmospheric rivers (ARs) in the UK is a mature, deep quasi-stationary low-pressure system near Iceland and an established high-pressure system over or near the Iberian Peninsula, the result being a strong southwesterly maritime flow from lower latitudes. In these situations, the distribution of the rainfall is likely to be more consistent with the position of the AR and clearly modulated over the western high ground. In contrast, users must note that the water vapour flux EFI will not produce a signal (indicative of large rainfall accumulations) from meteorological set-ups such as a slow-moving area of low pressure centred over England with persistent rain from a wrap-round occlusion.
The South Yorkshire flooding by the River Don on 7/8 November 2019 is an example where there was no EFI for water vapour flux signal in the run up to the event, although there was a weak EFI precipitation signal. Finally, there is a benefit in looking at the EFI for water vapour flux and precipitation at short range (days 1 to 3) as this can give the hydrometeorologist some confidence in the potential for an unusual event.
ECMWF and the FFC will continue to work together to collaborate on the use of this EFI product. The FFC feedback provided is likely to be added to the ECMWF Forecast User Guide and the suggested improvements will be discussed in future meetings.