Geoscience Reference
In-Depth Information
(WMO, 1986) showed, at that time, no clear advantage of the more complex models when comparisons
were made for a variety of catchments and over a number of years of data.
3.5 Distributing Meteorological Data within a Catchment
One of the problems in applying all these evapotranspiration and snowmelt models at the catchment scale
is taking account of the variability of meteorological conditions within the catchment. Insolation depends
on the angle and aspect of different slopes; wind speeds depend on wind direction and pressure gradients
in relation to the form of the topography; temperatures depend on elevation; humidities depend on
evapotranspiration upwind. Distributed rainfall-runoff models have the potential to take such variations
into account, but this requires a further model to distribute the meteorological data measured at one point
or, at best, a small number of points, to other points in the catchment using digital elevation and other
distributed data. This problem is greatest for hilly and mountainous catchments with a wide elevation
range. It is a particular problem for the estimation of snowmelt early in the season, since snowmelt
generally starts at lower elevations on slopes with a southerly aspect and may be significantly delayed at
higher elevations. Models for predicting distributed meteorological data within a catchment have been
suggested, for example, in the SAFRAN-CROCUS snowmelt model of Brun
et al.
(1993), in RHESSys
(Band
et al.
, 1991; Hartman
et al.
, 1999), and by Bl oschl
et al.
(1991).
Remember that there is also the problem in snowmelt modelling of knowing how much snow is there
to melt in the first place, since it is very difficult to obtain information on spatial patterns of snow depth
and density to get estimates of snow water equivalent. It is possible to use remote sensing to estimate
changing patterns of snow covered area which can be used as a constraint on snow melt models (e.g.
Bl oschl
et al.
, 1991; Rango, 1995).
3.6 Other Hydrological Variables
Rainfalls and discharge are the measured hydrological variables most often available to the modeller and
certainly are the most useful to the rainfall-runoff modeller. However, in some catchments, other types
of hydrological measurement may be available, such as measurements of standing water levels in wells,
profiles of soil moisture and spatial patterns of near surface soil moisture. Such data clearly gives more
information on the hydrological behaviour of a catchment but the amount of information may be limited
since, with the exception of a few research catchments, the number of measurement sites is likely to be
small. The scale of the measurements is also important in this context. Such internal measurements tend
to be small scale or “point” measurements, reflecting the hydrological conditions only in the immediate
vicinity and, to some extent, up-gradient. Thus, it may be difficult to compare such measurements with
the predictions of even the most distributed rainfall-runoff models available. Geophysical methods such
as ground-penetrating radar, electrical resistance tomography and cross-borehole tomography can give a
better indication of moisture patterns in space, but are still only relatively local in scale. The use of such
internal measurements in model calibration and evaluation are considered again in Sections 5.4 and 6.4.
3.7 Digital Elevation Data
In many developed countries of the world, digital elevation maps (DEM) or terrain maps (DTM) are
becoming available at a resolution fine enough to broadly represent the form of hillslopes (50 m in the
UK and France; 30 m in the USA; 25 m in Switzerland). DEMs with a fixed grid size are known as
raster
data. Digitised contour maps (
vector
DEMs) may also be available (Figure 3.3a). In fact, to date, most
