Modelling Catchment and Fluvial Processes and their Interactions - Environmental Modelling: Finding Simplicity in Complexity - page 182

(Vorosmarty et al ., 2000). Thus the complexities of soil
and vegetation effects at the small scale may be cancelled
out at the large scale where the interaction of rainfall
and geology (through topology dominate).

2012). The models range from simple 'black-box' repre-
sentations of input and output, which are often successful
in the prediction of runoff from rainfall, through to more
complex representations of some of the spatio-temporal
complexity of catchments. The latter are more capable
of fostering a better understanding of the reasons for
observed behaviour. Catchments themselves are superb
simplifiers (filters), converting a spatial complexity of
patterns and processes into a relatively simple and well
understood output - the hydrograph (see also Chapter 7).
The range of models available reflects the debate over the
most appropriate methods, the need to predict the out-
comes of specific interventions and scenarios for change,
the emphasis on explanation as well as prediction (for
the purposes of flood prediction and mitigation or water-
resource management) and the paucity of data available
for larger catchments compared with smaller ones.
The types of catchment model available include phys-
ically based models, based solidly on an understanding
of the physical processes, empirical models based on the
patterns observed in data and conceptual models that pay
little attention to the physics of the processes but, rather,
represent the catchment conceptually as, for example, a
series of cascading stores for water and the fluxes between
them. Models may be deterministic models in which a
given set of inputs will always produce the same output,
or stochastic models, which represent the variability of
parameters, processes or events using probability dis-
tributions and which thus attempt to handle some of
the inherent uncertainty in modelling and in data (see
Chapter 8). Models may be lumped at the catchment
scale, meaning that data and modelling are aggregated
at this scale, they may be lumped at the subcatchment
scale (and thus semi-distributed at the catchment scale)
or they may be fully distributed - that is, lumped at the
raster grid cell or TIN polygon scale.
Empirical models tend to be lumped, conceptual mod-
els tend to be semi-distributed and physically based
models tend to be fully distributed. The increase in
computing power and of available spatial data in the form
of remote sensing and GIS datasets, especially DEMs,
and remotely sensed imagery has vastly increased the
potential for distributed modelling. At the catchment
scale, to be based on physics, physically based models
have to be distributed and so 'distributed' and 'physically
based' often go together in catchment hydrological mod-
elling. Moreover many large catchments are ungauged
and thus cannot provide the calibration data necessary
for the development and parameterization of conceptual
or empirical models. A driving force for the development

Their climatic characteristics and their spatial variabil-
ity. The spatial distribution of temperature, radiation
and rainfall - which are themselves highly correlated
with elevation - will determine the spatial distribu-
tion of contributions from contributing areas within a
catchment.

•

Their vegetation cover and land-use characteristics
and its spatial variability. Chapter ten on hillslope
(eco-)hydrology indicates the significance of vegetation
and animals for the hydrological balance of hillslopes
and the partitioning of rainfall into local infiltration
and infiltration excess and thus runoff.

•

The spatial distribution of their human populations.
Populations are the source of demand for water and the
ecosystem services that it provides (see also Chapter 20).
The distribution of population may determine the
locations of local extractions and artificial storage of
water from the channel network or from locally gen-
erated runoff (as in the aljibes of North Africa and the
Mediterranean - van Wesemael et al ., 1998) and from
local groundwater sources. The location of populations
will also determine the magnitude of local land-use
change (see Chapter 18) with corresponding impacts
and the sources of point and nonpoint agricultural,
industrial and domestic pollution to the water courses.

•

Because of surface and subsurface lateral flows, hydro-
logical catchments are highly connected such that a
change in any one part of them can have implications for
a number of other parts downstream. Furthermore, the
spatial distribution of all of these factors relative to the
direction of streamflow is particularly important because
it determines the potential for cumulation or diminution
of streamflow along the flow path. A series of source areas
in line with the flow path will cumulate outflow along the
flowline whereas a mixture of source and sink areas will
tend to maintain or dampen outflow along the flow path.

11.2.4 Abrief reviewof catchmenthydrological
modelling

The use of models to understand better and to predict the
behaviour of water in catchments has a long history and,
because models are rarely successful in application outside
the catchments for which they were developed, there are
many models to be found (for a critical review, see Beven,