PUB in practice: case studies - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 287

is shown schematically in Figure 11.17 , consists of three
HRUs. Because of the practice of cereal crop rotation,
HRUs 1 and 2 alternate between being fallow and cropped,
with the third HRU representing a grassed stream channel.
As the grass has a greater aerodynamic roughness height
than the winter stubble remaining on the cropped HRU
(and both are greater than that of the fallow), snow will be
blown from the fallow to the cropped HRU, and from both
to the grassed channel. The runoff from HRUs 1 and 2 will
always drain to HRU 3, from which it exits the basin.
The virtual basin model was applied in continuous simu-
lations across the Canadian Prairies over a multi-decadal
period. Whilst topographic characteristics were held con-
stant for the virtual basin, soil texture was permitted to vary
with the location modelled

a) Synthetic basin - small stream

b) CRHM Hydrological Response Units

in general there are more clay
soils in the eastern part of the region and more silty-loams
in the west. This variation in texture has an important
influence on runoff generation in the model. The model
was applied with meteorological inputs from a few high-
quality meteorological stations using observations of daily
precipitation, and hourly air temperature, humidity, rainfall
and wind speed. All data were obtained from the Data
Access Integration (DAI) portal ( http://loki.qc.ec.gc.ca/
DAI/ ) , which was created through the cooperation of a
number of organisations. As many of the meteorological
time series were discontinuous, owing to stations being
moved, it was necessary to construct hybrid time series
by combining data from several stations. In all cases the
stations had been moved only short distances and were
identified within the records as being at the same general
location (i.e., town).
Because of the very poor density of measured solar
radiation stations on the Canadian prairies, solar radiation
was reconstructed using the techniques developed by
Shook and Pomeroy ( 2011 ). The latitude was also permit-
ted to vary with the location modelled to allow accurate
calculation of the components of the radiation balance.
The stations and the prairie ecozone, as defined by
Marshall et al.( 1996 ), are shown in Figure 11.18 . The
model was spun-up by running it from 1 January 1960 to 1
January 1961. As there is typically no change in soil
moisture over the frozen winter period and because the
summer of 1959 was dry, the soil moisture at each location
was initially set to be 50% of saturation

-

Field 1

Field 2

Grass

Figure 11.17. Schematic diagram of Creighton Tributary virtual
basin (a) and CRHM hydrological response units (b). The dashed line
indicates blowing snow redistribution.

of observation (Gray et al., 1985 ). The basin is character-
ised by level open land with poor drainage and highland
with rolling topography; it is drained by a grassland ' cou-

lée (sharp incised valley in the upland plain) from which

flows Creighton Tributary. This stream flows intermit-
tently, with most flow during and immediately after the
snowmelt period. Runoff on Creighton Tributary was
monitored from the 1960s to mid 1980s. Pomeroy et al.
( 2007b ) showed that CRHM, with an appropriate
model structure, could provide an excellent representation
of runoff, snowpack and snowmelt timing in Creighton
Tributary, without calibration of model parameters.
A virtual Creighton Tributary model is applied to represent
ungauged basin flow throughout the prairie region.

'

this is the only
state variable that required spin-up to estimate.
All comparisons were made relative to values calculated
from the climate normal period (1961

-

90) for each site
modelled. The models were run from 1961 (allowing for
one year of spin-up) through 1990, and the empirical
cumulative distribution function (ECDF) was computed
from the modelled annual discharges for each site. Having
determined the normal period ECDF for each site, the
exceedance probabilities were then determined for the

-

Virtual first-order basin regional simulation
To show how the hydrology of a basin modelled on
Creighton Tributary varies with drought and wet cycles
and with regional climatic variations, a virtual basin model
was constructed based on that used by Pomeroy et al.
( 2007b ) and Fang and Pomeroy ( 2007 ). The model, which