FLOOD FORECASTING MODELS AND UNCERTAINY REPRESENTATION - Modelling Uncertainty in Flood Forecasting Systems - page 10

Chapter 2
FLOOD FORECASTING MODELS AND
UNCERTAINY REPRESENTATION
Summary of Chapter 2
This chapter describes different types of models used for flood
forecasting as (i) physically-based, (ii) conceptual and empirical, and
(iii) data-driven; and presents a discussion regarding the uncertainty
assessment in these models. This chapter also defines uncertainty in rather
general terms, distinguishing 'types of uncertainty' from 'sources
of uncertainty'. Various types of uncertainty are discussed, such as
ambiguity (one-to-many relationships), randomness and vagueness
(or fuzziness), inherent and epistemic and quantitative and qualitative .
Specific to flood forecasting, four sources of uncertainty are identified:
input uncertainty, parameter uncertainty, model uncertainty and
natural and operational uncertainty. Widely used theories of uncertainty
representation are reviewed. The principles and introduction of prob-
ability theory, fuzzy set theory and possibility theory are presented.
2.1 Types of flood forecasting models
A conceptual structure of a typical flood forecasting, warning and response system
(FFWRS) is presented in Chapter 1 (Fig. 1.3). The flood forecasting and flood warning
sub-systems of the FFWRS are also defined in Subsection 1.2.2. A modelling system for
real time flood forecasting consists of one or more of the following components: (i) a
model of rainfall forecasting, (ii) a model of rainfall-runoff forecasting, and (iii) a model
of flood routing and inundation (see, for example, Reed, 1984 and Yu and Tseng, 1996).
Various aspects of flood modelling are presented by Price (2000). With the rapid
development in hydrological and hydraulic modelling techniques and the necessary
computer systems (both hardware and software), the trend these days is towards a
so-called integrated real-time flood forecasting system, which consists of all three
components of flood forecasting. It should however be noticed that in many parts of the
world less sophisticated techniques of flood forecasting and ad-hoc warning systems are
still in use.
The present study covers uncertainty analysis only in the rainfall-runoff and routing
components of flood forecasting systems. Therefore, discussion regarding the models of
rainfall forecasting is beyond the scope of this thesis. The model of rainfall-runoff
forecasting also consists of a number of components depending on the level of detail used
in the models. Here it is intended to cast some light on different types of available models
for flood forecasting and a discussion vis-à-vis their uncertainty assessment. These
models are commonly categorised on the basis of the extent to which they represent the