Drought impacts on Prairie land surface
hydrological dynamics
Kevin Shook, John W. Pomeroy, and
Rob Armstrong
Centre for Hydrology, University of Saskatchewan,
117 Science Place, Saskatoon, Saskatchewan, Canada, S7N 5C8
Prairie hydrology of 1999-2005
drought




Hydrology of the Canadian prairies is unique
Dominated by small streams and wetlands
which are ungauged
Evaluation of hydrological effects of 1999-2005
drought will require modelling
Ordinary models are useless for prairie
conditions
Future climate scenarios


Hydrological effects of future climate are also
unknown in prairies
Models developed for this project will also be
used to determine effects of future climate
changes on prairie hydrology
Canadian Prairie Hydrology






Wind redistribution of snow to wetlands and stream channels in
winter is critical to formation of runoff contributing areas.
Frozen soils enhance runoff efficiency during spring snowmelt
so that >80% of runoff occurs during snowmelt period.
Contributing area small and variable due to post glacial
topography, large depressional storage potential and lack of a
well developed fluvial drainage network.
Baseflow from groundwater often nonexistent due to heavy
glacial till substrate.
Drainage of small streams and wetlands ceases completely in
summer when actual evapotranspiration consumes most
available water.
Prairie streams are almost completely ungauged and often
altered by dams, drainage, water transfers, etc.
Prairie Runoff Generation
Snow Redistribution to Channels
Spring melt and runoff
Dry non-contributing areas to runoff
Water Storage in Wetlands
Modelling Prairie Hydrology




Need a physical basis to calculate the effects of
changing climate, land use, wetland drainage
Need to incorporate key prairie hydrology
processes: snow redistribution, frozen soils,
spring runoff, wetland fill and spill, noncontributing areas
Hydrological models developed elsewhere do not
have these features and fail in this environment
Streamflow calibration does not provide
information on basin non-contributing areas and
is not suitable for change analysis
Cold Regions Hydrological Model
Platform: CRHM




Modular – purpose built from C++ modules
Modules based upon +45 years of prairie hydrology research at
Univ of Saskatchewan
No provision for calibration or optimization, parameters set by
knowledge
Hydrological Response Unit (HRU) basis







landscape unit with characteristic hydrological processes
single parameter set
horizontal interaction along flow cascade matrix
Model tracks state variables and flows for HRU
HRUs assumed to represent one response type, basis for
coupled energy and mass balance
HRUs connected aerodynamically for blowing snow and via
dynamic drainage networks for streamflow
Incorporate wetlands directly using fill and spill algorithm
“Virtual” basins

The spatial variability of the drought will be
simulated using models of two typical prairie
basins:
1.a small upland stream, and
2.a small wetland complex


The virtual basins will be simulated all over the
praries, wherever sets of forcing variables are
available
Outputs during the drought period will be
compared to the climate normal period of 19611990
CRHM model of small prairie
stream basin
Small stream basin
CRHM model
HRU1


HRUs 1 and 2 alternate
between cropped and
fallow
HRU3 is grassed
HRU2
HRU3
CRHM model of small prairie
wetland
Small wetland complex
CRHM model
3
4
HRU 2
1
2
HRU 1


HRU 3
HRU 1 alternates between
cropped and fallow
HRUs 2-4 are wetlands
HRU 4
CRHM data requirements

CRHM requires only a few variables:

Daily


Hourly


Snowfall
Rainfall, Air temperature, RH,Windspeed,
Solar radiation
Solar radiation required for modelling snow
melt, evaporation and other processes

Currently, measurements are only available at
Regina
Simulated Data


NARR

Available as daily or 3-hour values

32 km grid

All required variables available, since 1979

Free!
but


Canadian precipitation data not assimilated by
NARR
Windspeeds and precipitation are unusable
Estimation of Qsi



Incoming short-wave radiation (Qs i) can be
calculated directly if the atmospheric
transmittence is known
Many researchers have found simple empirical
relationships between daily atmospheric
transmittance and the range of daily air
temperatures (ΔT)
Given the daily estimated Qs i, CRHM can
calculate the hourly components of the solar
radiation (incoming, outgoing, shortwave and
longwave)
Annandale Daily Qsi




Annandale method
yields daily Qsi
values about as good
as NARR’s
Effects of scatter
even out over a
season
Very simple to
calculate
Now built into CRHM
Sites
Brandon HRU1 SWE
250
200
SWE (mm)
150
100
50
0
Jan-15 Feb-12 Mar-11 Apr-8 May-6 Jun-3 Jul-1 Jul-29 Aug-26 Sep-23 Oct-21 Nov-18 Dec-16
Jan-1 Jan-29 Feb-26 Mar-25 Apr-22 May-20 Jun-17 Jul-15 Aug-12 Sep-9 Oct-7 Nov-4 Dec-2 Dec-30
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Gridded simulation results


Drought period values divided by normals for
ease of comparison
Resulting values gridded (thin-plate spline) to
show spatial variability
1999 Annual Discharge/Normal
2000 Annual Discharge/Normal
2001 Annual Discharge/Normal
2002 Annual Discharge/Normal
2003 Annual Discharge/Normal
2004 Annual Discharge/Normal
2005 Annual Discharge/Normal
Summary




Simulations have been run at all 15 locations
using the first CRHM model
Simulation results are preliminary
The spatial and temporal extents of the
hydrological drought can now be visualised
Second CRHM model has yet to be used
Acknowlegements

SGI Canada

DRI

Data Access Integration
( http://quebec.ccsn.ca/DAI/) for data
Apart from CRHM, this research was done
entirely with Open Source Software:

R
- data reduction, statistics

Qgis, SAGA - GIS

Open Office - presentation