Retrieval of snow wetness Martti Hallikainen Contributions from: Norut, IFAc, NR

1
Retrieval of snow wetnessMartti Hallikainen
Contributions from Norut, IFAc, NR
2
Snow wetness by SAR

• Parameter Snow wetness (SW)
• Definition Percentage of water in snow pack
  (range 0-15)
• State of the art Presence of wet snow can easily
  be detected by change detection against a dry
  snow reference scene
• Theory Qualitative measurement of SW is
  challenging since model inversion is involved.
  For single polarization, single frequency SAR,
  there are too many unknowns. Using combinations
  of change detection against summer scenes and
  assumptions about surface roughness may allow
  inversion.

3
EnviSnow development

• Norut Detection of wet snow by change detection
  uncalibrated SW product showing grades of snow
  wetness
• IFAC Detection of snow wetness with (a) SAR
  using threshold and time series and (b) microwave
  radiometry
• NR Retrieval of snow wetness from optical data
  using temperature and snow grain information

4
Norut SW produt
Presence of wet snow is detected for each
acquisition and delivered as a standard product
in near real time. The product may be used as
extra information to hydrologists and hydropower
industry
20050524
20050405
5
Norut Graded SW-product
Graded snow wetness products have also been
produced
Wet
Wetness is based on backscatter difference
between reference scene and wet snow scene and
arbitrary scaled
Dry
Valdresflya/Jotunheimen, 2004
6
IFAC Mapping wet snow cover in alpine areas
Italy
Backscattering coefficient (dB)
Backscattering coefficient (dB)
Backscattering coefficient (dB)
7
Multitemporal Analysis 2004

• Wet snow detected with a threshold set to 3 dB
• Dry snow in April Pixels classified as wet snow
  in May but not in April.
• Snow free in May Pixels classified as wet in
  April but not in May
• Dry snow at both dates Pixels not classified as
  wet snow at both dates, but located at a height
  above the median altitude of the wet snow in
  April

5 April 2004
10 may 2004
Light blue dry snow Blue wet snow Green
forests Brown bare soil Red layover and shadow
areas (rocks)
8
Sensitivity of microwave emission to snow wetness
Measurements on Cordevole, Italy
Experimental and simulated brightness
temperatures at 19 and 37 GHz (vertical
polarization) together with simulated volumetric
liquid water content of the two snow layers as a
function of time.
Tb at 37 GHz and 19 GHz ,V pol. as a function of
LWC of the upper and lower layer of snowpack.
9
NR Retrieval of snow wetness from optical data

• Objective Infer approximate snow wetness from
  the development of snow parameters in a time
  series of optical images
• Idea Snow temperature close to 0C combined with
  a rapid increase of the effective snow grain size
  is a strong indication of snowmelt start
• Approach Combine snow temperature (STS), snow
  grain size (SGS) and snow cover (SCA)
• STS for today
• Recent time series of SGS
• SCA to mask out lt 100 SCA areas
• Cloud mask

Field measured snow temperature, and satellite
measured snow temperature and effective snow
grain size for Heimdalen test site in 2003. HH
Heimdalshø, VF Valdresflya
10
Some results
White - dry, cold snow STS lt -2C.
Light/dark blue - dry/moist -2C lt
STS and -0.5C Yellow/orange - moist -0.5C
lt STS 0.5C. Red - wet 0.5C lt 1.0C
Unchanged SGS. Increasing SGS
11
Conclusions (1/2)

• Snow wetness cannot easily be derived from single
  pol/frequency SAR without assumptions or a priori
  knowledge (e.g. from in-situ measurements)
• Simple quantitative SW products indicating
  presence of wet snow and grades of snow wetness
  may be produced operationally, and may be useful
  in hydrology
• Microwave radiometer output is highly sensitive
  to snow wetness several frequencies can be
  employed

12
Conclusions (2/2)

• Snow temperature and effective grain size can be
  accurately measured using optical sensors
• Method limited to pixels of 100 SCA
• STS and SGS very sensitive to presence of
  snow-free surfaces
• Snowmelt onset time clearly identifiable
• Coarse wetness classes feasible