FLOOD FORECASTING MODELS AND UNCERTAINY REPRESENTATION - Modelling Uncertainty in Flood Forecasting Systems - page 11

physics of the processes involved. On this basis the flood forecasting models can be
classified as
1. Physically-based
2. Conceptual and empirical
3. Data-driven or black box.
2.1.1 Physically-based models
The physically based models, as the name suggests, are based on the mathematical
representation of all pertinent physical processes. The European Hydrological System or
SHE (Abbott et al. 1986a & b) is an example of a systematic approach for the
development of a physically based, fully distributed, catchment modelling system. SHE is
physically based in the sense that the hydrological processes of water movement are
modelled, either by finite difference representations of the partial differential equations of
mass, momentum and energy conservation, or by empirical equations derived from
independent experimental research (Abbott et al., 1986b). In fact, no models of a
complex natural phenomenon like floods can be exclusively represented by mathematical
equations and there exists some degree of approximation and empiricism. In this sense,
even a sophisticated modelling system like SHE is not a complete representation of the
physics of the natural phenomenon. However, the level of physical representation in SHE
is such that it has been widely recognised as a standard physically based catchment
modelling system.
Application of a distributed, physically-based model such as the SHE requires the
provision of large amounts of parametric and input data, some of which may be time
dependent (Abbott et al. 1986b). Such data are often unavailable. Problems may also
arise due to an inadequate process representation in building a physically based model,
for example, the representation of macropore flow as matrix flow. Moreover, such models
handle large arrays of data and involve iterative solutions, which require considerable
computing time and resources. Therefore, despite their theoretical advantages, engineers
tend to rely more on conceptual and/or empirical models in many practical applications.
There are however, some situations where physically based models can be very useful
and should be preferred, for example, to assess the impact of changes in the physics of
the catchment or for the assessment of extreme events (Guinot and Gourbesville, 2003).
2.1.2 Conceptual and empirical models
A conceptual model is generally composed of a number of interconnected conceptual
elements. Conceptual models are not fully physically based, but their developments are
inspired by the (limited) understanding of the physical processes. In other words these
models endeavour to incorporate some aspects of the physical processes with some
simplifications. There are many conceptual models with different levels of physical
representation that have been widely used in hydrological forecasting. Fleming (1975)
presented a thorough review of several conceptual models. Crawford and Linsley (1966)
are credited for the development of the first major conceptual model by introducing the
well known Stanford Watershed Model (SWM), which has undergone many
modifications thereafter (Shamseldin, 2002). Some other widely used conceptual models