Using ECMWF's Forecasts (UEF2019)

The deterministic approach to weather prediction does not allow to
establish a-priori the degree of skill of an individual forecast;
instead, probabilistic forecasts provide a more complete, reliable and
accurate view of what could happen in the future, ideally providing
information on the relative frequency of an event occurring.
Therefore, they bring definite benefits for decision-makers.
Forecast users can exploit such information, for example, when they
want to weight the losses associated with adverse weather events
against the costs of taking precautionary actions.
The aim of this work is to assess the added value of enhanced
horizontal resolution in the probabilistic prediction of upper-level
and surface fields.
In particular, the performances of three different ensemble systems
were compared: ECMWF-ENS (51 members, 18 km horizontal resolution),
COSMO-LEPS (20 members, 7 km horizontal resolution) and COSMO-2I-EPS
(20 members, 2.2 km horizontal resolution).
While the first 2 ensemble systems are operational, COSMO-2I-EPS is
still in a development phase.
The intercomparison window covers two separate periods, characterised
by different weather types:
- the former one (20-27 June 2016) presents convective precipitation
events with weak synoptic forcing,
- the latter one (15 October-15 November 2018) is mainly dominated by
large-scale forcing, with stratiform precipitation.

In both cases, high-impact weather events affected different areas of Italy.

In this work, both upper-level and surface variables are analysed.
As for the surface, 2-metre temperature and precipitation cumulated
over six hours were verified against the non-conventional station
network provided by the National Civil Protection Department.
The ensemble spread and the root mean square error of 2-metre
temperature were computed, while a number of probabilistic scores
(Brier Skill Score, Ranked Probability Score, ROC-Area, Outliers
Percentage and others) were considered for precipitation.
The best scores were mainly obtained by the COSMO-based ensemble
systems which have higher horizontal resolution and lower ensemble
size; in particular, the newly implemented COSMO-2I-EPS often achieved
the best performances.

Although these results are based over two relatively short periods,
they show the added value of high resolution in mesoscale ensembles,
which turn out to be more skillful in the probabilistic prediction
of atmospheric fields at all levels.