Flood forecast

Question 1

Predictability

Answer 1

 Assessment is possible by comparing lead time TL with time of concentration Tc and flood wave translation time Tw

 As more the catchment is storing water as later the peak will occure and as easier is the forecast.

 The flood peak will occure latest after the time of concentration Tc

 Predictability declines with decreasing catchments size: a) because of smaller time of concentration and b) because of higher dynamic of flood producing rainfall

Question 2

There are three basic cases:

Answer 2

Case 1: TL>Tw
Rainfall Runoff model and precip. observation with remote sensing

Case 2: TL(Tc+Tw)
Rainfall Runoff model with precipitation forecast

Question 3

Methods

Answer 3

Flood routing models (short term):

• Hydrological (e.g. Muskingum)
• Hydraulic (e.g. Saint-Venant)

• Conceptual (e.g. bucket type models)
• System hydrological (e.g. Einheitsganglinie)

• Time series models (ARMA)
• Softcomputing methods

Question 4

Model updating (1)

Answer 4

Utilisation of observation of target variables (e.g. discharge) and/or observation of status variables (like soil moisture) for an updating of the forecast model.

Question 5

Model updating (2)

Answer 5

 Model updating makes forecasting more complex but can increase the prediction performance

 Makes especially sense if the forecast has systematic errors or the errors are autocorrelated

 There are different methods for updating: e.g.

a) simple updating assuming persistence of errors
b) updating with stochastic prediction of errors
c) Kalman-Filter technics (see literature)