Inferring snow

Towards a new representation of snow metamorphism
in macroscopic snowpack models
[email protected]
C. M. Carmagnola1, S. Morin1, F. Dominé2
1Météo-France/CNRS,
CNRM-GAME URA 1357, Centre d’Etudes de la Neige, Grenoble, France, 2CNRS / Université Joseph Fourier Grenoble 1, LGGE UMR 5183, Grenoble, France.
Context:
• Snow on the ground is a complex porous
medium that dramatically modifies the energy
balance of the surface and the chemical
composition of the overlying atmosphere.
• Its impact on climate and atmospheric chemistry
is determined in part by the snow physical
properties, which in turn depend on the size of
snow grains, their shapes, their spacing and the
strength of their interconnections.
Goal:
Perspectives:
Replace the Crocus semi-quantitative description of the snow grains along
metamorphism (dendricity and sphericity) by the incorporation of new
quantitative variables (SSA, kT) as state variables describing the physical
and microstructural properties of snow.
• Inferring snow types from ρ, SSA and kT (transform SSA and kT from diagnostic
to prognostic variables of the model).
Fresh snow
Fragmented snow
Candidate state variables: SSA and kT
•
Rounded grains (ET metam)
Rounded grains (wind metam)
Faceted grains
Depth hoar
Specific surface area (SSA):
Melt clusters
- quantifies the surface/mass ratio of snow (expressed in m2 kg-1)
Fresh snow
- related to the optical radius (r) through: SSA =
• From snowfall to snowmelt, the progressive
transformation of snow illustrates the fact that
snow is also a dynamic medium whose
properties evolve over time through snow
metamorphism.
3
r ⋅ ρice
- measurable (NIR reflectance, micro-CT)
Fragmented
snow
Rounded grain s
(ET
metamorphism)
Rounded
grains
Faceted
grains
Depth hoar
Melt cluster
150-400
(wind
metamorphism)
ρ
ρ (kg m-3 )
10-200
60-250
150-350
300-600
130-300
150-300
SSA (m2 kg-1)
35-160
15-90
11-30
15-38
8-45
7-22
2-5
kT (W m-1 K-1)
0.03-0.12
0.06-0.2
0.08-0.3
0.2-0.6
0.08
0.03-0.15
0.1-0.5
Dominé et al., ACP, 8, 174-208, 2008.
• Is it enough? Do other variables depend only on ρ, kT and SSA?
The current snowpack model: CROCUS
DUFISSS:
• The detailed snowpack model Crocus was developed at CNRM-GAME/CEN in
the 1980s.
- based on a semi-quantitative description of the time
evolution of the morphological properties of the snow
grains along with snow metamorphism
- one-dimensional
model
multilayer
SSA vs density for different types of snow grains (from
Dominé et al., 2008).
thermodynamic-based
•
- energy and mass balance of the snowpack
• Snow grains are described by their sphericity and dendricity as long as there is
fresh snow in the layer, otherwise they are described by their size and
sphericity.
Thermal conductivity (kT): F = −kT
This system measures the
NIR
reflectance
in
an
integrating sphere at 1310
and 1550 nm and allows
accurate and rapid SSA
measurements of a snow
sample.
• Derive physically-based laws to describe the time evolution of the new state
variables.
Micro-structure-based models
Experimental data analysis
…
- related to snow type, grain size and bonds between grains
Example of the time evolution of ρ (from a
simple newtonian densification law), kT
(from the CROCUS power law) and SSA
(from the parametrerisation of Tallandier
and Dominé, 2007); simulation within a
given layer over 140 days.
- effective value: heat conduction in the ice, heat conduction in the air, transfer
of latent heat
- measurable (heated-needle probe, micro-CT)
TG (temperature gradient)
dρ
dT


= f  t,T ,
, SSA, kT , LWC 
dt
dz


• Develop a new CROCUS version based on mesurable state variables.
dT
dz
MF (melt-freeze)
ET (equi-temperature)
dSSA
dT


, ρ , kT , LWC 
= f  t,T ,
dt
dz


dkT
dT


, ρ , SSA, LWC 
= f  t,T ,
dt
dz


Dendricity varies from 1 to 0 (decreasing
systematically); it corresponds
to the share of fresh snow
crystals present in a layer of
snow.
Sphericity varies from 0 to 1.
sphericity =
Acknowledgments:
rounded
rounded + angular
However, dendricity and sphericity cannot
be determined in the field and they are not
directly related to the relevant physical
properties of snow.
We thank B.Lesaffre, Y. Lejeune and J.-M. Willemet for useful discussions.
We thank also M. Flanner for providing his model to predict the evolution of dry snow SSA.
Heated-needle probe:
This system measures the temperature
rise of a heated metal wire inserted
into the medium: for a given heating
rate, the lower the thermal conductivity
the higher the resulting temperature
rise.
References:
kT vs density for different types of snow grains (from
Sturm et al., 1997). The CROCUS model uses the
1.88
 ρ sn ow 
relationship:

k sn ow = kice
 ρ ice 
Brun et al., JG, 35(121), 333-342, 1989 and JG, 38(128), 13-22, 1992.
Dominé et al., JGR, 112, F02031, 2007.
Dominé et al., ACP, 8, 174-208, 2008.
Flanner and Zender, JGR, 111, D12208, 2006.
Legagneux and Dominé, JGR, 110, F04044, 2005.
Sturm et al., JG, 43, 143, 1997.
Taillandier and Dominé, JGR, 112, F03003, 2007.