Castaldini, Doriano
modelling for real-time forecasting and warning as well as the development of scenarios for climatic change impact
assessment.
Despite the effort invested in flood control works world wide, neither flood occurrences nor damages are decreasing
(Kundzewicz & Takeuchi, 1999). In fact, recent floods have not only exceeded previous damage record (due to
increasing urban population and confidence in flood protection measures), but also absolute record stages.
Problems exist with respect to the understanding of the process itself and calibration. Causes are frequently associated
with extreme precipitation, steep slopes and poorly infiltrating shallow soils (e.g. Peschke, 1998). Samuels (1998)
names precipitation, structure failure and marine conditions as caused. Gilvaer & Black (1999) name overtopping as the
most common mechanism that results in flood embankment failure. However, the knowledge of the origins of floods and
its possible magnitude in a given region remains unclear (Bérod et al., 1999).
These models must be calibrated and validated on observed data, but extreme events, land-use change and possibly
climatic fluctuations will typically be outside recorded experience (Peschke, 1998). The floods that took place recently
in the Panaro River basin occurred under different circumstances than floods recorded in earlier centuries. The course of
the Panaro river has been modified in the last centuries (Pellegrini ef al., 1979; Castaldini & Piacente, 1999) and large
records referring to equal conditions are not likely to exist. This problem increases when predictions are made with
respect to future conditions or when for current conditions no large data records exist.
The establishment of a more disaster conscious society with improved preparedness (safe-fail) rather than unrealistic
fail-safe (safe from failure) design of flood defences seems more sustainable (Kundzewicz & Takeuchi, 1999). Safe-fail
means that when a system fails, it fails in a safe way. Flood defence does not guarantee complete protection.
Communication of risks in a tangible way will increase consciousness among the public more when using living with
floods as a guiding principle rather than prevention. Samuels (1998) notes that with respect to flood management return
periods are not a helpful way of communicating risk to the public. Therefore, the goal of the model presented here is not
predicting when a flood will occur, but rather prediction of the consequences of a flood event of a given magnitude.
4.1 Model structure
The model assumes relevant processes to take
place at different locations and model variables
are a therefor function of space. The processes
and their parameters also vary in time, and
these variations can be incorporated, by
computing updated values for variables (e.g.
quantities of water) after an arbitrarily chosen
amount of time has elapsed. The model can
thus be called a distributed dynamic model.
The basis of the model is formed by the basic
water budget, where the change in storage
volume (here regarded as the floodwater on the
surface) is determined by inflow and outflow of
water from a unit area. Inflow will be consist
mainly of precipitation, surface runoff from the
upstream area and spilled water from the
channel. Processes such as subsurface drainage
and base-flow respond much slower,
particularly in areas with small gradients.
Outflow will consist mainly of surface runoff
(back into channel or to another unit area) an
infiltration.
Spatial input data for the model is stored in
maps representing topography (for derivation
of drainage network and slope), hydraulic
conductivity, storage capacity and a map
depicting the presence of infrastructure (plus
alternative railway routes) (Fig. 4). The storage
capacity map is prepared on the basis of
precipitation records of the period preceding the
Figure 4. Input and structure of the model
232 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.
