Operational European Flood Forecasting - Environmental Modelling: Finding Simplicity in Complexity - page 412

Figure 25.7 EFAS 'probability' map for the
forecast of 5 January 2011. 12:00 (From
EFAS).

25%, 50% and 75%. Critical thresholds are plotted as
shaded areas in the corresponding threshold colour so
that visually it is clear at one glance which threshold has
been exceeded. However, perhaps most usefully, EFAS
also synthesizes the results in daily information for indi-
vidual hotspot locations, which aids the decision-maker
in synthesizing information and assessing consistency of
information between different forecasts.
Figure 25.8 is an example of a typical box diagram
from EFAS. For each day and each type of forecast, the
highest threshold exceeded is visualized, together with
the gradient of the hydrographs to rise or fall illustrated
by up and down arrows. Peaks are marked as stars. For
EPS only the number of EPS exceeding the five-year
return period threshold is shown. The example shown
in Figure 25.8 illustrates this representation for a river-
gauging station in the Mosel, a tributary of the Rhine.
Clearly, EFAS forecasts based on deterministic forecasts
of DWD (seven-day forecast from 5 to 11) and ECMWF
(ten-day forecast from 5 to 14) only exceed the EFAS
medium alert threshold with a peak on the 9 January.
However, more than half of all EPS exceed this threshold

on the 8 and 9 January (EPS > HAL), whereas the 20-year
return period threshold (SAL) is not exceeded. The event
is forecast at the end of the lead time of the COSMO-LEPS
forecast (5 to 9).
This representation was chosen because it allows the
user to visualize easily persistence in time of the individual
forecasts. If the forecasts persistently show similar results
from forecast to forecast, it is likely that the situation is
captured well by the meteorological models and thus the
results can be trusted more than a random distribution
of exceedances. This can be the case, for example, for
convective situations where generally unstable meteoro-
logical conditionsmay indicate some widespread flooding
in a particular region, but in fact the river basins affected
change from day to day.
A comprehensive description of all EFAS products is
available on the EFAS website http://floods.jrc.ec.europa
.eu/efas-flood-forecasts.html and in Ramos et al . (2007).
In addition, further examples of EFAS flood forecasts can
be found in Thielen et al (2008) where a case study of the
Romanian floods of October 2007 are presented.