Geoscience Reference
In-Depth Information
•
Are the runoff coefficients consistent for increasing storm volumes (allowing for seasonal variations)?
For example, are any runoff coefficients greater than 100%? This would indicate that one or other of
the measurements is in error since mass balance makes it difficult for a catchment to produce more
runoff outputs than rainfall inputs.
•
If more than one discharge gauge or raingauge are available, check for consistency between the gauges
(normalising for differences in area for discharges). Compare runoff coefficients or use
double mass
curves
to check for changes in slope in the accumulated volumes at different gauges.
•
Check for any obvious signs that infilling of missing data has taken place. A common example is
where measured rainfall intensity is apparently constant for a period of 24 hours, suggesting that a
volume from a daily raingauge has been used to fill in a period where the recording raingauge was not
working. Hydrographs with long flat tops are also often a sign that there has been a problem with the
measurements.
These types of simple check are easy to make and at least allow some periods of data with apparently
unusual behaviour to be checked more carefully or eliminated from the analysis. There is a danger, of
course, of rejecting periods of data on the basis that a chosen model cannot be made to give a good
simulation of that period. Unless there is some other reason for rejection, this should not be considered
good practice, since it is normally the case that the modeller learns more about the limitations of a model
from situations where it cannot give good simulations than where it does. The reader should remain
aware, however, that most of the hydrological modelling literature tends (quite naturally) to report the
best simulations with any given model rather than the worst!!
3.3 Meteorological Data and the Estimation of Interception
and Evapotranspiration
3.3.1 Estimating Potential Evapotranspiration
In many environments, evapotranspiration makes up a larger proportion of the catchment water balance
than stream discharge. Thus, for longer periods of rainfall-runoff simulation, it will generally be necessary
to estimate actual evapotranspiration from a catchment in order to have an adequate representation of the
antecedent state of the catchment prior to each rainfall event.
We must distinguish here between estimates of
potential evapotranspiration
and
actual evapotran-
spiration
. Potential evapotranspiration is the loss expected over a surface with no limitation of water.
It is a function of the
atmospheric demand
, that is the rate at which the resulting water vapour can be
moved away from the surface. The atmospheric demand depends primarily on the energy available to
convert liquid water to vapour from net radiation, the humidity gradient in the lower atmosphere, the
wind speed, and the roughness of the surface. Rough surfaces, such as forests, will have higher potential
evapotranspiration rates than smooth surfaces, such as a lake, given similar radiation, humidity and wind
conditions. In general, actual evapotranspiration rates will be at the potential rate until the water supply
from the soil becomes limiting.
Methods for estimating potential evapotranspiration range from using a simple annual sine curve to
the more physics-based Penman-Monteith equation detailed in Box 3.1. A simple seasonal sine curve for
daily potential evapotranspiration, regardless of the variations in the weather, would appear to be far too
simplistic to be a useful model of potential evapotranspiration. The study of Calder
et al.
(1983), however,
showed that such a curve could give results equally as good as more complex formulations requiring
more data in modelling soil moisture deficits for several sites in the UK. They treated the mean daily
potential evapotranspiration, the only parameter required by the model, as a parameter to be calibrated
and found that similar values could be used for all their study sites. A seasonal sine curve can be defined
