Improving hydrological model representation of
alpine snowcover depletion and snowmelt runoff
generation
Chris DeBeer and John Pomeroy
Centre for Hydrology
University of Saskatchewan
Saskatoon, Canada
Background
• Late winter alpine snowcover has high spatial variability
in depth, energy state, and SWE
• Together with spatial variation in melt energy, this leads
to patchy snowcover and variable runoff contribution
• Important to represent this variability in hydrological
models and land surface schemes
• Disaggregating terrain into slope and aspect based units
can resolve much of the variability for modelling
Background
• Approach is to apply slope-corrected melt rates to SWE
distribution on each landscape unit to derive snowcover
depletion curve
0.02
0
0
75
150
SWE (mm)
225
Ma
p(SWE)dSWE
1
300
40 mm/d
225
150
SCA Fraction
p(SWE)
0.04
∫
Accumulated Melt (mm)
Relative Frequency
SCA =
∞
20 mm/d
75
10 mm/d
0
SCD curve
0.75
0.5
0.25
0
0
2
4
6
8
Days
10
12
14
0
2
4
6
8
Days
10
12
14
Background
• However, melt timing and rates are non-uniform, even
within a landscape unit
• Important to consider internal processes in the snowpack
when applying the energy balance principle in physically
based models
– e.g., internal energy, refreezing overnight
– Depends on depth and density, so varies over SWE distribution
• Oversimplifying to assume uniform melt and runoff
generation over a distribution of SWE
Objectives
• Examine small scale variation in melt timing and rates
due to variation in SWE depth in a sparsely vegetated
alpine environment
• Determine the effect of this variation on simulated
snowcover depletion and meltwater runoff generation
• Suggest an approach for improving the representation of
small scale snowmelt variability in hydrological models
Study Site
•
Fisera Ridge and Upper Middle Creek Basin
Areal Snowcover Observations
Time lapse digital photography used to
monitor areal snowcover depletion
May 13,
June
July
1,
4,
9,
10,
14,
17,
19,
22,
26,
29,
31,
7,
2,2007
4,
7,
10,
13,
18,
21,
24,
27,
2007
2007
2007
2007
2007
Snowmelt Modelling and Validation
• Point location snowmelt modelling and observation
– snow depth and temperature monitoring on slopes
– Melt rates simulated using Snobal module
in Cold Regions Hydrological Model
(Qm = LVE + H - K↑ + K↓ + L↓ - L↑ + G - dU/dt)
LvE
H
K↑
K↓
L↓
L↑
Soil layer
E
U
Active layer
Lower layer
P
U
G
R
Snowmelt Modelling and Validation
South-east facing site
North facing site
Simulated depth (m)
2
1
0
1-Apr
15-Apr
29-Apr 13-May 27-May 10-Jun
Simulated depth (m)
0.5
0
1-Apr
24-Jun
Simulated SWE (mm)
Measured SWE (mm)
600
300
0
1-Apr
15-Apr
29-Apr 13-May 27-May 10-Jun
24-Jun
15-Apr
29-Apr 13-May 27-May 10-Jun
Date (2008)
24-Jun
Snow temp (°C)
Snow temp (°C)
Measured active layer T
Simulated lower layer T
Measured lower layer T
-25
1-Apr
24-Jun
Simulated SWE (mm)
Measured SWE (mm)
100
0
1-Apr
15-Apr 29-Apr 13-May 27-May 10-Jun 24-Jun
Date (2008)
Simulated active layer T
-15
29-Apr 13-May 27-May 10-Jun
200
Date (2008)
-5
15-Apr
Date (2008)
Depth (mm)
Depth (mm)
Date (2008)
900
Measured depth (m)
1
Depth (m)
Depth (m)
Measured depth (m)
-5
Simulated active layer T
Measured active layer T
-15
-25
1-Apr
Simulated lower layer T
Measured lower layer T
15-Apr 29-Apr 13-May 27-May 10-Jun 24-Jun
Date (2008)
Simulated Snowmelt Variation
South-east facing site
North facing site
Time-dependent
small scale
associations
between melt rate
and SWE due to
more rapid
warming of
shallow snow.
Associations
change later in
melt period as all
snow is ripe.
Effects on SCD and Runoff Contributing Area
South facing slope
•
•
SWE distributions measured
by surveys and repeat LiDAR
SCD simulated by applying
both uniform melt and inhomogeneous melt for
different SWE classes
Snowmelt runoff contributing
area defined as fraction of
basin generating runoff > 5
mm/d
1
Area fraction
•
Simulated Melt rates used to
predict SCD in adjacent cirque
basin
0.8
simulated
SCD (In)
0.6
simulated
SCD (U)
0.4
observed
SCD
0.2
SRCA
0
r
r
r
r
y
ay
ay
Ap
Ap
Ap
Ap - Ma
M
M
2
4
6
24
26
28
30
Date (2008)
North facing slope
1
Area fraction
•
0.8
simulated
SCD (In)
0.6
simulated
SCD (U)
0.4
observed
SCD
0.2
SRCA
0
r
r
r
r
ay
ay
ay
Ap
Ap
Ap
Ap
M
M
M
2
4
6
24
28
26
30
Date (2008)
Effects on SCD and Runoff Contributing Area
•
Spatial distributions of
snow depth from
LiDAR used to map
spatial patterns of
SCD and SRCA
•
Differences in melt
and SWE distributions
control the spatial
development of the
snowmelt runoff
contributing area
Effects on SCD and Runoff Contributing Area
1
0.8
Area fraction
South
facing
slope
simulated
SCD (In)
simulated
SCD (U)
0.6
0.4
observed
SCD
0.2
SRCA
0
r
Ap
24
ay
M
1
ay
M
8
-M
15
ay
-M
22
ay
-M
29
ay
n
Ju
5-
un
-J
2
1
un
-J
9
1
un
-J
6
2
l
Ju
3
Date (2008)
1
0.8
Area fraction
North
facing
slope
simulated
SCD (In)
simulated
SCD (U)
0.6
0.4
observed
SCD
0.2
SRCA
0
r
Ap
24
ay
M
1
ay
M
8
-M
15
ay
-M
22
ay
-M
29
ay
n
Ju
5-
Date (2008)
un
-J
2
1
un
-J
9
1
un
-J
6
2
l
Ju
3
Effects on SCD and Runoff Contributing Area
• Results from each slope were aggregated to derive basin
scale snowcover depletion and runoff contributing area
1
simulated
SCD (In)
Area fraction
0.8
simulated
SCD (U)
0.6
0.4
observed
SCD
0.2
SRCA
0
r
Ap
24
ay
M
1-
ay
M
8-
-M
15
ay
-M
22
ay
-M
29
ay
n
Ju
5-
Date (2008)
n
Ju
12
n
Ju
19
n
Ju
26
l
Ju
3
Snowmelt Runoff Simulation
• This framework can be applied to simulate the meltwater
inputs over the basin
– Inputs to each slope computed by running the model in point
mode for various initial SWE depth classes over the distribution
– Simulated outflow from the base of the pack weighted by the
fraction of the total distribution in each SWE class, and
aggregated on each slope unit
Snowmelt Runoff Simulation
– Rate and timing of melt input over the basin is sensitive to
approach, with significant implications for the snowmelt
hydrograph
400
300
lumped basin and
uniform melt
0.4
lumped basin and
inhomogeneous melt
3
500
0.5
distributed basin and
inhomogeneous melt
0.3
measured hydrograph
0.2
200
0.1
100
0
1-Apr-09
Discharge rate (m /s)
Cumulative melt input (mm)a
600
0
21-Apr-09 11-May-09 31-May-09 20-Jun-09 10-Jul-09
Date
30-Jul-09
Conclusions
• Snowmelt timing and rate is non-uniform, even within
individual landscape units, due to differences in the
internal energetics and ripening of the snow
• Effect of initial SWE depth on melt rates is pronounced in
early melt period with major implications for areal SCD
• Simulations of areal snowcover depletion and snowmelt
runoff generation can be improved by considering
separate SWE classes on individual slopes
• By improving the representation of melt inputs over the
landscape, the snowmelt hydrograph can likely be more
accurately simulated