IMPRINTS: IMproving Preparedness and RIsk maNagemenT for flash floods and debriS flow events- EC FP7 project
|
|
Gämschtobel 20 June 2007 (Foto: Chr. Rickli, WSL)
|
IMPRINTS aims to contribute to a better
preparedness and a better operational risk management for flash floods and
debris flow events. These improvements shall help reducing fatalities and
economic damage, caused by these kinds of natural hazards.
To achieve this methods and tools are developed
which will help risk managers and decision makers in emergency agencies and
utility companies to take the necessary measures in good time.
Background
To do flood
forecasting a lot of information is needed. On the one hand the hydrological
model has to know about the properties of the respective catchment as topography,
land use and soil properties. The hydrometeorological state of the catchment
has an influence on the future runoff behaviour of the catchment. If the soil
is already saturated an additional rainfall event will contribute immediately
to the runoff. But to forecast future runoff it’s not only the past weather
conditions but also the current weather forecast which is of importance.
The weather
forecasts provide a certain framework to the hydrological model of the probable
special and temporal distribution of rainfall events and their intensity in the
near future.
The problem
We all experienced
that we should better not relay 100 % on weather forecasts because they are
governed by numerous uncertainties. The weather forecasts are derived from
atmospheric models. But as the current state of the atmosphere cannot be
described exactly the model is run several times with slightly different
initial conditions. The result of these runs is an ensemble of possible weather
scenarios. The ensembles serve as input to the hydrological model which itself
produces an ensemble of possible runoff. Such ensemble forecasts are called
probabilistic forecasts. The uncertainties arising form the description of the
state of the atmosphere propagate through the model chain and increase. But
because local thunder storms are very difficult to forecast with such coarse
atmospheric models radar rainfall forecasts are used to get short-term
forecasts, so called nowcasts. Those are more likely to detect and pursuit thunder
cells and help for a more local rainfall estimation.
Other
uncertainties incorporated in a semi-distributed hydrological model arise form
the assumptions made about the state of the catchment and from the
parameterization of the model itself.
To do
The aim is
a better quantification of these uncertainties and through that to increase the
accuracy of flood forecasts. They should be able to give information about the
probability of the different scenarios and so to simplify the planning and
decision making process of utility companies and emergency agencies.
Models used
Test beds in Switzerland
The project
is very practice-oriented. So the future end-users of the methods and tools to
be developed are involved. In Switzerland this is the department for
construction and environment of the canton Glarus, the Verzasca SA and the
Azienda Elettrica Ticinese with the corresponding test beds Linth down to the
gauge at Mollis and Verzasca down to the gauge at Lavertezzo.
Contact
