ICESat AMSR-E_ model

Document Sample
ICESat AMSR-E_ model Powered By Docstoc
					Impact of freeboard and snow depth
  variabilities on ICESat sea ice
        thickness retrievals

Nathan Kurtz, Thorsten Markus, Lynn Sparling,
                Don Cavalieri
       Combining freeboard and snow depth to get
                   sea ice thickness

                                               ●   ICESat freeboard retrievals
hs = snow depth                AMSR-E, model       represent only a thin line through
hi = ice thickness    ICESat
hf = freeboard                                     a generally larger scale snow
                                                   depth grid cell (12.5 km x 12.5
                                                   km and larger)

                                               ●   Need to understand how spatial
                                                   variability of snow depth affects
                                                   ice thickness retrieval error
                 hs                      s
hf                                             ●   How does the resolution of snow
            hi                          i         depth affect the accuracy of
                                                   ICESat retrieved thickness
                                   w              values?
                              Data sets
●   March 21, 22, and 25 areas used,
    flights cover 50 km x 100 km
    with ~5 km spacing between
    flight lines

●   High resolution airborne data:
    ATM laser altimeter
    measurements converted into
    freeboard, PSR microwave
    radiometer data converted into
    snow depth

●   High resolution aircraft data are
    used to simulate low resolution
    satellite data
         –   How representative are
             thin lines of a larger area?

                i   i
                    h   h
    ●   Distribution of delta is
        approximately Gaussian
    ●   Standard deviation of delta
        gives the error in assuming the
        mean of the altimeter track is
        equal to the mean of the larger
        snow depth pixel area
Spatial variability results
              ●   Variability of snow-ice freeboard
                  decreases with pixel size except for
                  mixed mixed first-year and multi-year
                  ice region
              ●   Snow depth variability is low for all
                  pixel sizes
              ●   Spatial variability errors found to vary as
                  a percentage of the mean snow depth of
                  the area
              ●   For AMSR-E, using several snow depth
                  pixels and a longer ICESat transect
                  reduces spatial variability error
              ●   Effect on ice thickness retrieval error for
                  the study areas: average of 24 cm, with
                  reductions to 12 cm possible by using a
                  longer measurement transect
         Resolution of ice thickness
●   How does the resolution of the snow depth affect
    the accuracy of ICESat retrieved thickness
●   High resolution method: scale snow depth
    measurements to the ICESat scale to get the sea
    ice thickness distribution
●   The ice thickness distribution is important for
    calculations of oceanic heat flux, with thin ice
    dominating the heat exchange process
   Freeboard and snow depth relation

Common feature in aircraft snow depth vs. freeboard plots: a linear
increase in snow depth with increasing freeboard; then randomly
distributed around a common mean
 The point where snow depth and
 freeboard are no longer correlated can
 be estimated from the large-scale mean
 snow depth and freeboard of the area,
 regression of the data gives:

 fbcutoff 0.69 hs   0.22   fb   5.10

 hsthick 1.03 hs    .83

Once the above values are found, the slope of the linear portion can be
●   Since snow depth
    variability is low, we can
    assume hsAMSR hsi N
●   Iterative adjustment of
    hs_thick used to conserve
    the snow depth
●   Results compare well
    with distributions derived
    from full resolution
    aircraft data
        ICESat first year Arctic sea ice
           thickness distributions
February-March 2006      ●   Combinations of AMSR-E
                             snow depths with ICESat
                             freeboards can be used to
                             provide ice thickness values
                             for first year ice

October-November 2005    ●   Scaling the snow depth to the
                             resolution of ICESat is found
                             to be necessary to resolve the
                             distribution of thin ice
●   The contribution of spatial variability in the retrieval
    error is small and can be predicted from the mean snow
    depth of the area
●   A method to scale the snow depth to the resolution of
    ICESat has been developed and can be used to resolve
    the sea ice thickness distribution
●   Better estimates of ocean-atmosphere heat flux by using
    the ice thickness distribution

Shared By: