Water Management Implications of
CRHM: Mountains to Prairies to Arctic
John Pomeroy and colleagues
Centre for Hydrology,
University of Saskatchewan, Saskatoon
Problem
Rocky Mountain Water Resources are Declining
WHY?
Mean Yearly Flow m3/sec
60
Bow River at Banff, Mean Annual Flow
50
40
30
20
10
Trend -11.5% since 1910, Statistically Significant at 99% Probability
0
1900
1920
1940
1960
Year
1980
2000
2020
Flows in Late Summer
Late summer flows large and dropping rapidly
Mean August Flow m3/sec
120
Bow River at Banff, Mean August Flow
100
80
60
40
20
0
1900
Trend -24.9% since 1910, Statistically Significant at 99.9% Probability
1920
1940
1960
Year
1980
2000
2020
Early Spring Flow Increasing
Winter flows small and rising somewhat
12
Mean March Flow m3/sec
Bow River at Banff, Mean March Flow
10
8
6
4
2
Trend +15.2% since 1910, Statistically Significant at 99.9% Probability
0
1900
1920
1940
1960
Year
1980
2000
2020
Marmot Creek Research Basin
•
•
•
•
•
•
•
1450-2886 m.a.s.l. Kananaskis Valley, Bow River
Alpine
Subalpine
Montane
Clearcut
Meadow
900 mm
Marmot Basin
precipitation
• 70% snowfall
• ~50% runoff
y
Rive
kis
s
a
an
Ka n
lle
r va
w
Bo
y
lle
va
r
e
Riv
Temperature Trends at High Elevation in
Marmot Creek, Rocky Mountains
Winters are warmer by 3 to 4 oC since 1962
Harder & Pomeroy
Upper Clearing
• 1844 m
• Small forest
clearing
• Sheltered by fir
and spruce forest
Fisera Ridge, Mt Allan Cirque
• 2318 m
• Alpine
Ridge
• Windblown
CRHM for Alpine Terrain
Meteorological Inputs T, RH, U, P, K↓
Latitude, elevation, slope,
aspect, vegetation, fetch, area
Distributed K↓, L↓, u*, T, q, snowfall, rain
Albedo Decay
Blowing Snow Model ΔSWE, Sublimation, Transport
Energy Balance Snowmelt Model
Melt, Sublimation
SWE Variability
Snow Covered Area Depletion Model
Alpine Ridge Snow Modelling
snow depth (m)
0.7
Fisera Ridge
0.6
observed
snow depth
0.5
calculated
snow depth
0.4
0.3
0.2
0.1
0
01/11/06
01/12/06
31/12/06
30/01/07
01/03/07
31/03/07
rms error = 0.134 m, mean error = 0.074 m
30/04/07
30/05/07
• Dominant windflow: north to south
• Flow over ridgetop and into forest
Windflow
Sublimation Loss
Forest
South Face (bottom)
South Face
(top)
Downwind Transport
Ridge
Top
North Face
700
600
RMSE = 9.7 mm
MB = 0.00
SWE (mm)
500
400
300
200
100
0
24-Sep-08
24-Oct-08
24-Nov-08
24-Dec-08
24-Jan-09
24-Feb-09
24-Mar-09
Snowfall
NF SWE (SIM)
Ridge SWE (SIM)
SF Upper SWE (SIM)
SF Lower SWE (SIM)
Forest SWE (SIM)
NF (OBS)
Ridge (OBS)
SF Upper (OBS)
SF Lower (OBS)
Forest (OBS)
SF top
Transport Out
Ridgetop
NF
Transport Out/Snowfall
Transect
100%
75%
50%
25%
0%
Forest
SF bottom
SF top
Transport In
Ridgetop
NF
Transport In/Snowfall
Transect
200
150
100
50
0
50%
40%
30%
20%
10%
0%
Forest
SF bottom
SF top
Blowing Snow Sublimation
Ridgetop
NF
Transect
Sublimation/Snowfall
Transport Out/Snowfall
SF bottom
400
300
200
100
0
Blowing Snow Sublimation (mm)
Transport In (mm)
Forest
Transport In/Snowfall
50%
40%
30%
20%
10%
0%
Blowing Snow Sublimation/
Snowfall
Transport Out(mm)
200
150
100
50
0
Winter Warming Impact on
Alpine Ridge Snow Accumulation
200
reference simulation
150
+ 1°C
+ 2 °C
SWE (mm)
+ 3°C
100
+ 4 °C
50
0
30/11/06
30/12/06
29/01/07
28/02/07
30/03/07
29/04/07
29/05/07
Impact of Warming on Blowing
Snow Fluxes
Cumulative Blowing Snow Transport or Sublimation (mm)
160
150
140
130
Transport
Sublimation
120
110
100
90
80
0.0
+0.5
+1.0
+1.5
+2.0
+2.5
Temperature Increase ( ºC)
+3.0
+3.5
+4.0
maximum SWE (mm)
Impact of Winter Warming on
Maximum Snow Accumulation
200
150
100
50
0
0
1
2
3
air temperature increase (°C)
4
Impact of Winter Warming on
Snowmelt Rate
melt rate (mm SWE /day)
9
8
7
6
5
4
3
2
1
0
0
1
2
temperature increase (°C)
3
4
Impact of Winter Warming on
Spring Snowmelt Duration
35
duration of spring melt (days)
30
25
20
15
10
5
0
0
1
2
temperature increase (°C)
3
4
Impact of Winter Warming on Date
of Snowpack Depletion
temperature increase (°C)
4
3.5
3
2.5
2
1.5
1
0.5
0
03-May
08-May
13-May
18-May
23-May
date of snowcover depletion
28-May
02-Jun
CRHM for Mountain Forests
Forest Snow Modelling
0.7
snow depth (m)
0.6
Upper Clearing
observed snow
depth
calculated snow
depth
0.5
0.4
0.3
0.2
0.1
0
01/11/06
01/12/06
31/12/06
30/01/07
01/03/07
31/03/07
30/04/07
30/05/07
Winter Warming Impact on
Mountain Forest Snow Regime
Snow Accumulation (mm)
120
100
Current
+1 C
80
+2 C
+3 C
60
+4C
40
20
0
10/10/06 09/11/06 09/12/06 08/01/07 07/02/07 09/03/07 08/04/07 08/05/07 07/06/07 07/07/07
250
Snowfall
Snowmelt
Current Climate
Forest Snow Processes
mm water equivalent
200
Sublimation
Snow Accumulation
150
100
50
0
10/10/06 09/11/06 09/12/06 08/01/07 07/02/07 09/03/07 08/04/07 08/05/07 07/06/07 07/07/07
250
Snowfall
+4 C Climate
Forest Snow Processes
mm water equivalent
200
Snowmelt
Sublimation
Snow Accumulation
150
100
50
0
10/10/06
09/11/06
09/12/06
08/01/07
07/02/07
09/03/07
08/04/07
08/05/07
07/06/07
07/07/07
Change in Melt with Temperature
Cumulative Snowmelt (mm)
160
140
Current
+0.5 C
+1 C
120
+1.5 C
100
+2 C
80
+3 C
60
+3.5 C
40
+2.5 C
+4C
20
0
10/10/06 09/11/06 09/12/06 08/01/07 07/02/07 09/03/07 08/04/07 08/05/07 07/06/07 07/07/07
Change in Snowfall with
Temperature
Seasonal Snowfall (mm)
250
200
150
100
50
0
Current
+0.5 C
+1 C
+1.5 C
+2 C
+2.5 C
+3 C
+3.5 C
+4C
Change in Sublimation with
Temperature
Intercepted Snow Sublimation (mm)
80
70
60
50
40
30
20
10
0
Current
+0.5 C
+1 C
+1.5 C
+2 C
+2.5 C
+3 C
+3.5 C
+4C
Change in Maximum Accumulation
with Temperature
Maximum Snow Accumulation (mm)
120
100
80
60
40
20
0
Current
+0.5 C
+1 C
+1.5 C
+2 C
+2.5 C
+3 C
+3.5 C
+4C
Water Management Implications
• These results show that intact mountain forests
have a mitigating effect on some aspects of
climate variability and that wind-swept open
environments are highly sensitive to climate
warming.
• Full consideration of blowing and intercepted
snow processes along with energy balance
snowmelt calculations must be given for credible
climate change impact studies of mountain snow
hydrology.
Effects of Forest Cover Change
• Evergreen forest canopy is associated with
two primary hydrological effects
– Snow and rainfall interception and subsequent
sublimation and evaporation resulting in
reduced sub-canopy snowfall or rainfall,
– Alteration of sub-canopy radiation and
turbulent transfer affecting the snowmelt rate.
Forest Sky View for Maximum Melt
Energy from Net Radiation
1.0
0.8
south-sloping site (αs= 0.7)
south-sloping site (αs= 0.8)
level site (αs= 0.7)
v[]
0.6
level site (αs= 0.8)
north-sloping site (αs= 0.7)
0.4
north-sloping site (αs= 0.8)
0.2
0.0
2/1/07
3/1/07
4/1/07
5/1/07
6/1/07
There is no simple relationship between forest density and melt rate.
Influence of slope, aspect, solar elevation, weather and albedo are overwhelming.
Ellis and Pomeroy, in preparation
CRHM Forest
Observed and Modelled Forest
Energetics – Marmot Creek
CRHM Forest Tests –
Colorado, Switzerland, Alberta
Slope and Forest Density Effect on Net
Radiation for Snowmelt - Rockies
Clearing
Mature Forest
Net radiation = solar + thermal radiation
Ellis &
Pomeroy,
in
preparation
Water Management Implications
• Forest clearing increases snow
accumulation
• Forest clearing accelerates snowmelt rates
on south facing slopes and level sites, BUT
• Forest clearing reduces snowmelt rates on
north facing slopes
Prairie Runoff Generation
Snow Redistribution to Channels
Spring melt and runoff
Dry non-contributing areas to runoff
Water Storage in Wetlands
PRAIRIE HYDROLOGY – Limited
Contributing Areas for Streamflow
Localized hydrology
affected by poor drainage,
storage in small
depressions
Non-contributing areas for streamflow
extensive in Canadian Prairies
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, non-contributing areas
• Frustration that 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
Smith Creek Hydrology Study
•
Problem: Inability to reliably model the basins of the
Upper Assiniboine River and other prairie basins where
variable contributing area, wetlands, nonsaturated
evapotranspiration, frozen soils, snow redistribution
and snowmelt play a major role in hydrology.
•
Objectives
–
–
–
Develop a Prairie Hydrological Model computer program that
can simulate the response of streams, wetlands, and soil
moisture to weather inputs for various basin types.
Evaluate the model performance in Smith Creek by comparing
to observations of streamflow, wetland extent, and snowpack.
Use the Prairie Hydrological Model to estimate the sensitivity
of streamflow, wetland water storage, and soil moisture to
changes in drainage and land use.
Smith Creek – extreme interannual and
seasonal variability
Smith Creek, Saskatchewan, ~400 km2 basin area
Streamflow m 3 per second
25
20
Average 1975-2006
15
1995 High Year
10
2000 Low Year
5
0
27-Dec
27-Nov
28-Oct
28-Sep
29-Aug
30-Jul
30-Jun
31-May
01-May
01-Apr
02-Mar
31-Jan
01-Jan
Streamflow over Time
Top 4 years for streamflow since 1995
30000000
Smith Creek Annual Streamflow
3
Annual Streamflow (m )
25000000
20000000
15000000
10000000
5000000
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
1975
0
Peak Flow over Time
Maximum Daily Discharge of Smith Creek during 1975-2006
Top 6 peak daily flows since 1995
3
Maximum Daily Discharge (m /s)
25
20
15
10
5
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
1975
0
Changing Climate?
Mean Annual Temperature (°C)
5
Mean Annual Air Temperature at Yorkton
4
Warming but high variability
3
2
1
y = 0.0239x + 1.4259
R2 = 0.0404
0
600
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
1975
-1
A n n u a l R a in fa ll a n d S n o w fa ll a t Yo rk to n
Ra in f a ll ( mm)
500
S n o w f a ll ( mm)
400
300
200
100
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
0
1975
No trend in
rainfall and snowfall
Drainage of Wetlands?
Drainage of Wetlands?
Modelling Approach
CRHM – Prairie Hydrological Model Configuration
Flow Chart in Cold Regions
Hydrological Model Platform
(CRHM)
HRU Configuration for Smith Creek
HRUs “grouped”
into “representative
basins”, RBs, that
are repeated for
sub-basins but with
individual parameter
sets. Routing
between RBs
permits large scale
process estimation.
Small scale Processes
Large Scale Processes
Routing
Fallow
Stubble
RB 1
Grassland
RB 2
Woodland
RB 3
Wetland
Open
Water
River
Channel
(a)
Amongst HRU in a
Representative Basin
RB 5
RB outlet
Smith Creek
basin outlet
RB 4
(b)
Amongst Representative Basins
Instrumentation of Smith Creek
Completed
Summer 2007
Hydrometeorological Station
11 dual rain gauges
7 wetland level recorders
Main Hydrometeorological Station
Temperature, humidity, wind speed,
shortwave radiation,
longwave radiation, soil moisture,
soil temperature,
soil heat flux, snow depth, rainfall,
snowfall
Snow and Wetland Surveys
Smith Creek Basin Characteristics
Drainage Network
Spot Image
Remote Sensing Supervised Classification
SPOT5
Field tests of
vegetation
classification
vegetation
used to define
HRU area,
HRU location &
vegetation
parameters
LT-2/SCR-4
SCR-8
LR-5
SCR-9
LiDAR-Derived DEM Drainage Network
LiDAR DEM to Calculate
Depression Storage using
pond volume-depth-area
relationship
Derivation of Wetland Depressions
CRHM Tests Smith Creek – No Calibration
Obse rv e d SWE v s Simulate d SWE at Smith Cre e k Sub-basin 1
Snow Accumulation
(mm SWE)
300
250
Fallow Obs. SWE
Fallow Sim. SWE
Channel Obs. SWE
Channel Sim. SWE
Wetland Obs. SWE
Wetland Sim. SWE
200
150
100
50
0
7-Feb
18-Feb
29-Feb
11-Mar
22-Mar
2-Apr
13-Apr
2008
Volume tric Soil M oisture at Smith Cre e k during Spring Snowme lt
Pe riod
Volumetric Soil
Moisture
0.5
0.4
Observed
Simulated
0.3
0.2
0.1
0
22-Mar
31-Mar
9-Apr
18-Apr
2008
27-Apr
6-May
Runoff Prediction 2008
Smith Creek Spring Discharge near Marchwell
Daily Mean Discharge (m 3 /s)
5
4.5
4
3.5
Observation
Non-LiDAR Simulation
LiDAR-based Simulation
3
2.5
2
1.5
1
0.5
0
22-Mar
27-Mar
01-Apr
06-Apr
11-Apr
16-Apr
21-Apr
26-Apr
01-May 06-May
2008
Non-LiDAR Simulation
LiDAR-based Simulation
Observation
MB
-0.07
-0.39
RMSD (m 3/sPeak Discharge (m 3/s)
0.10
4.61
0.12
4.17
4.65
Runoff Prediction 2009
Smith Creek Spring Discharge near Marchwell
Daily Mean Discharge (m 3 /s)
9
Observation
8
Non-LiDAR Simulation
7
LiDAR-based Simulation
6
5
4
3
2
1
0
23-Mar
28-Mar
02-Apr
07-Apr
12-Apr
17-Apr
22-Apr
27-Apr
02-May 07-May
2009
Non-LiDAR Simulation
LiDAR-based Simulation
Observation
MB
-0.21
-0.57
RMSD (m 3/Peak Discharge (m 3/s)
0.28
7.83
0.31
5.37
6.22
45.00
"normal condition"
40.00
complete forest cover
35.00
primarily agricultural land use
30.00
high wetland extent
25.00
minimal wetland extent
20.00
15.00
10.00
5.00
2004-05
2003-04
2002-03
2001-02
2000-01
1999-2000
1998-99
1997-98
1996-97
1995-96
1994-95
0.00
1993-94
Spring Discharge (1000 dam3)
Sensitivity Analysis: Change in
Spring Discharge
Sensitivity of Spring Discharge
Volume to Land use and Drainage
Change in discharge volume (1000
dam3)
6
Agricultural Conversion
Forest Conversion
4
Wetland Restoration
Wetland Drainage
2
0
0
5
10
15
20
25
-2
-4
-6
-8
-10
Spring Discharge Volum e (1000 dam 3)
30
35
40
% Change in Discharge Volume
Long-term Impact of Land Use and
Drainage Change
70
60
50
40
30
20
10
0
-10
-20
-30
-40
Forest
Conversion
Agricultural
Conversion
Wetland
Restoration
Wetland Drainage
Wetland Change in Low Discharge
Volume Year
Daily Mean Discharge (m 3 /s)
Scenarios of Smith Creek Spring Discharge near Marchwell
1.6
"Normal Condition"
1.4
high natural wetland extent
1.2
minimum wetland extent
1
0.8
0.6
0.4
0.2
0
11-Feb
20-Feb
29-Feb
09-Mar
18-Mar
27-Mar
05-Apr
14-Apr
23-Apr
02-May
2000
2000 Drought: Lowest Discharge Volume on Record
Wetland Change in High Discharge
Volume Year
Daily Mean Discharge (m 3 /s)
Scenarios of Smith Creek Spring Discharge near Marchwell
30
25
20
"Normal Condition"
high natural wetland extent
minimum wetland extent
15
10
5
0
01-Mar 10-Mar 19-Mar 28-Mar 06-Apr
15-Apr
24-Apr 03-May 12-May 21-May 30-May
1995
1995 Flood: Record High Discharge Volume
Discussion on Scenarios
• Changes in wetland extent often are accompanied by
changes to land use.
• Increasing forest cover decreases discharge volume.
• Increasing agricultural land increases discharge volume.
• Increasing wetland area reduces discharge volume, whilst
decreasing wetland area results in an increase.
• The changes to discharge volume due to decreasing
wetland area are similar for almost all discharge volumes,
but changes due to increasing wetland area tend to increase
with discharge volume.
• In dry conditions, when storage is small, wetland drainage
increases discharge volume, whilst wetland restoration has
little impact.
• In flooding conditions, when storage is filled, neither
wetland drainage nor restoration has an effect on the
hydrograph.
Conclusions
• Consideration of snow, frozen soil and surface storage
processes are essential to calculating spring runoff in the
Prairies.
• Depressional storage is exceedingly difficult to calculate in
this flat, poorly drained environment – LiDAR permits
estimation of depressional and wetland storage volumes.
• It is possible to model prairie snowpack, soil moisture and
streamflow without calibration using physically based
simulations that aggregate landscape scale hydrological cycle
calculations, if high resolution information is available on
catchment characteristics.
• There is moderate sensitivity of streamflow volumes to
changes in agricultural and forest land use.
• There is strong sensitivity of streamflow volumes to wetland
drainage and restoration.