358
G.A. Schultz
then transfonn the rainfall hyetographs into flood hydrographs. Remote sensing
data are used for both, the estimation of parameters of such models and as model
input. An important issue is the fact, that flood forecasting and control can be
achieved efficiently only, if the leadtime from the moment of forecast until the
occurance ofthe event (e.g. a flood peak) is reasonably long. This means, that the
relevant data (remote sensing and conventional) have to be available in real-time
at the forecasting and control center and, while the flood is still in progress realtime forecasts have to be made rapidly and frequently in order to achieve a long
leadtime and make the flood management measures efficient.
Since also snowmelt often plays an important role in flood generation, snowmelt
runoff models based on remote sensing infonnation are useful in flood forecasting.
This topic is dealt with in Chap. 11.
In this chapter only two types of remote sensing data are used, i.e. groundbased
weather radar and several types of geostationary and polar orbiting satellites.
16.2 General Approach
16.2.1 Modeling Philosophy
As mentioned above flood hydro graphs have to be derived from remote sensing
data with the aid of mathematical hydrological models. This has to be done in
real-time in the fonn of flood forecasts.
Model Types. For flood forecasting purposes usually two different modeling
approaches can be chosen: Stochastic models and detenninistic models. The latter
may be sub-divided into lumped and distributed system models. Although stochastic models can be used for forecasting purposes they will not be discussed
here, since they are usually not amenable to remote sensing data. Both types of
detenninistic models, lumped as well as distributed may be used in connection
with remote sensing data. More efficient use of the area distributed REMOTE
SENSING data is made by distributed models. The application of this model type
will be presented along with an example in Sect. 16.3.
Flood Forecasting Procedure. The transfonnation of remote sensing data into a
forecast flood hydro graph may follow a procedure consisting of four consecutive
steps: (1) acquistion of remote sensing data, (2) transfonnation of these data into
hydrometeorologically relevant infonnation (model parameter estimation as well
as model input generation), (3) transfonnation of rainfall (or/and snowmelt) computed under (2) into a runoff hydro graph and (4) the use of such forecast runoff
hydrographs for real-time control of flood protection measures, e.g. reservoirs.
The Forecasting Problem. Compared to conventional flood hindcasting procedures (computation of floods, which occurred previously with the aid of rainfallrunoff models) the forecasting problem is much more complex due to the facts,
that (a) at the beginning of a flood not enough relevant infonnation is available
and that (b) the whole forecasting procedure has to be done in real-time rather
G.A. Schultz
then transfonn the rainfall hyetographs into flood hydrographs. Remote sensing
data are used for both, the estimation of parameters of such models and as model
input. An important issue is the fact, that flood forecasting and control can be
achieved efficiently only, if the leadtime from the moment of forecast until the
occurance ofthe event (e.g. a flood peak) is reasonably long. This means, that the
relevant data (remote sensing and conventional) have to be available in real-time
at the forecasting and control center and, while the flood is still in progress realtime forecasts have to be made rapidly and frequently in order to achieve a long
leadtime and make the flood management measures efficient.
Since also snowmelt often plays an important role in flood generation, snowmelt
runoff models based on remote sensing infonnation are useful in flood forecasting.
This topic is dealt with in Chap. 11.
In this chapter only two types of remote sensing data are used, i.e. groundbased
weather radar and several types of geostationary and polar orbiting satellites.
16.2 General Approach
16.2.1 Modeling Philosophy
As mentioned above flood hydro graphs have to be derived from remote sensing
data with the aid of mathematical hydrological models. This has to be done in
real-time in the fonn of flood forecasts.
Model Types. For flood forecasting purposes usually two different modeling
approaches can be chosen: Stochastic models and detenninistic models. The latter
may be sub-divided into lumped and distributed system models. Although stochastic models can be used for forecasting purposes they will not be discussed
here, since they are usually not amenable to remote sensing data. Both types of
detenninistic models, lumped as well as distributed may be used in connection
with remote sensing data. More efficient use of the area distributed REMOTE
SENSING data is made by distributed models. The application of this model type
will be presented along with an example in Sect. 16.3.
Flood Forecasting Procedure. The transfonnation of remote sensing data into a
forecast flood hydro graph may follow a procedure consisting of four consecutive
steps: (1) acquistion of remote sensing data, (2) transfonnation of these data into
hydrometeorologically relevant infonnation (model parameter estimation as well
as model input generation), (3) transfonnation of rainfall (or/and snowmelt) computed under (2) into a runoff hydro graph and (4) the use of such forecast runoff
hydrographs for real-time control of flood protection measures, e.g. reservoirs.
The Forecasting Problem. Compared to conventional flood hindcasting procedures (computation of floods, which occurred previously with the aid of rainfallrunoff models) the forecasting problem is much more complex due to the facts,
that (a) at the beginning of a flood not enough relevant infonnation is available
and that (b) the whole forecasting procedure has to be done in real-time rather