Snow and glacier change modelling
in the French Alps
M. Dumont, M. Lafaysse, S. Morin, V. Vionnet (CNRM-GAME)
I. Zin (LTHE)
International Network for Alpine Research Catchment Hydrology
Inaugural Workshop
Barrier Lake Field Station, Kananaskis Country, Alberta, Canada
22-24 October 2015
General framework for global changes in the Alps
 Warming is expected to accelerate throughout the 21st century
 Seasonal shifts in precipitation and relative humidity are expected
 Precipitation and temperature extremes are expected to intensify
 Snow cover is expected to drastically decrease below 1500–2000m
 Changes related to droughts and natural hazards are expected
Some addressed modelling issues
 Model complexity vs. accuracy
 Dealing with uncertainties
 Remote sensing and local data assimilation
 Examples on the proposed INARCH & surrounding sites
The SURFEX/ISBA-CROCUS
soil-snow model
 1D model
 Multi-layers
max 50 couches
 Dynamical
discretization of
the layers
 Métamorphosis
of snow grains
Brun et al., 1989,1992; Vionnet et al., 2012
At local scale – Col de Porte
5
Model complexity vs accuracy – local scale
Same land surface model with:
 Same multi-layer
ground scheme
 3-layers vs multi-layers
snowpack scheme
 Small differences between
the various models for
bulk snow properties
 Strong impact of the input
meteorological data
Masson et al. 2013
Model complexity vs accuracy – local scale
JULES Investigation Model, 4 years performance assessment
 Same multi-layer
ground scheme
 1701 combinations of
multilayer
parameterizations for
albedo, fresh snow
density, compaction,
turbulent exchanges,
snow cover fraction,
thermal conductivity
 Best models vary from
year to year
Essery et al. 2013
 Same kind of results on SWE and snow depth for experiments
with different DDF schemes
At catchment scale – Arve headwater
SAFRAN meteorological analysis
Incoming
Air temperature
Incoming
Rel. humidity Windspeed solar radiation longwave
radiation
Rainfall
North
South
Conceptual 2-buckets module
for groundwater storage
Snowfall
At catchment scale – Arve headwater
Elevation bands
(300m)
Exposition
classes
(by 45°)
Slope classes
(by 20°)
Glacierized
areas
50 m
250 m
+ soil and vegetation types
1 km
 553
HRUs on the Arve headwater8 km
At catchment scale – Arve headwater
At catchment scale – Arve headwater
Glacial retreat between 1986 (solid line)
et 2012 (dashed line)
 low impact on discharge
(consistent with observations)
Model complexity vs accuracy
headwater scale
Same surface model, with the same multi-layer ground scheme
3-layers snowpack scheme
multi-layers snowpack scheme
 Significant differences during the snowmelt season
(not expected, cf. Masson et al. 2013)
 Process (and parameter !) interaction effect at headwater scale
Dealing with uncertainties – driving met data
Arve headwater
12 years of ensemble precipitation predictions at catchment scale
NCEP-GEFS 1°x1°, 20 members + ctrl
ECMWF-ENS 0.25°x0.25° members + ctrl
CNR-OPALE-GFS 40 members
Bellier et al., to be submitted
 Good performance of ECMWF-ENS even at small scale ! 
 Analog-based techniques unbias and make large scale ensemble
predictions more reliable 
 Diurnal cycle of performance 
Dealing with uncertainties - Ensemble predictions
 PEARP-S2M
- 1st May 2015 in Mont-Blanc area
Exemples:
Chamonix
02/05/2015
T = 5 yrs
Dealing with uncertainties - Ensemble predictions
 PEARP-S2M
- 1st May 2015 in Mont-Blanc area
Exemples:
Chamonix
02/05/2015
T = 5 yrs
Dealing with uncertainties - Ensemble predictions
 PEARP-S2M
- 1st May 2015 in Mont-Blanc area
Exemples:
Chamonix
02/05/2015
T = 5 yrs
Remote sensing and snow data assimilation
MODIS
Control
simulation
Charrois et al., to be submitted
Remote sensing and snow data assimilation
Reflectances assimilation (Refl_DA)
Reduced envelopes
dispersion
Reduced uncertainty on the
ending melt date
Need of regular observations
Charrois et al., to be submitted
Remote sensing and snow data assimilation
Reflectances + snow depth assimilation
(Refl+SD_DA)
Observations assimilation
0,07
0,04
Control simulation
18,5
9,5
10 days
Charrois et al., to be submitted
Climate projections + uncertainties (Durance)
Mean annual temperature (2000m)
Winter precip (2000m)
Dark blue : GCM
uncertainty
Green : downscaling
uncertainty
Cyan : residual
Red : large scale
internal variability
Yellow : small scale
internal variability
Snow cover duration at 1650m
Snow cover duration at 2250m
Lafaysse et al., 2014