Byrd Polar Research Center Remote Sensing Lab

1
Byrd Polar Research CenterRemote Sensing Lab

• CReSIS Members
• K. Jezek
• K. Farness
• S. Mather
• B. Csatho (now at SUNY-Buffalo)
• B. ORourke
• K. Leibfacher (undergrad)
• S. Westfall (undergrad)

2
Digital Elevation Model
Processing and Products
3
Current Status and Objectives
Available DEMs of the Greenland Ice Sheet are
based on satellite radar altimetry data (e.g.,
Bamber and Ekholm, 2001), or on the combination
of radar altimetry data with AVHRR imagery by
using photoclinometry techniques to fill the
regions between radar orbits (Scambos and
Harran). Elevation errors of these DEMs range
from a few meters at the middle of the ice sheet
(gt2000 m) to 50-100m in marginal regions. Our
objective is to improve the accuracy of these
DEMs by using ICESat data. The following
example is shown for the boxed region, however a
whole ice sheet DEM is currently being
developed.
4
Input Data

• 3/20/03- 6/23/05 ICESat satellite laser altimetry
  observations (L1, L2a-c, L3a-c) (L3d and L3e
  should be added)
• Airborne Topographic Mapper laser altimetry data
  from 1999-2003 (Krabill, NASA WFF)
• Existing DEMs from Bamber et al. (2001) and
  Scambos and Harran
• Landcover map of Greenland from KMS
• Categories ice sheet, local ice caps, land and
  ocean

5
Processing Steps

• Step 1 Detection and removal of outliers from
  the ICESat and ATM data sets.
• Only points on ice sheet and local ice caps are
  used
• ICESat outliers elevations are compared with
  median elevation of 2 X 2 km grid cells. Points
  with elevation difference gt 40 m from the median
  are removed.
• ATM outliers are removed by using the same
  technique. In addition, all observations in
  cells with less than 2 data points are removed.
• Step 2 Residuals between the ICESat data and the
  photoclinometry DEM are computed at each ICESat
  point over the ice sheet and local ice caps.
  Residuals are set zero at grid posts over land
  and ocean
• Step 3 Residuals are interpolated into a regular
  grid by using minimum curvature interpolation.

6
Processing Steps cont.

• Step 4 Improved DEM Available DEM
  interpolated residual
• Step 5 The accuracy of the new DEM is assessed
  by using the ICESS data set (compressed ATM
  airborne laser scanning data derived by plane
  fitting to original laser point swaths)

7
Distribution of Data Points ICESat (red) and ATM
(blue)
8
Photoclinometry DEM (Upper Panel) Interpolated
ICESat Residual (Lower)
Merged DEM from ICESat Data and
Photoclinometry DEM
9
Close Up Comparison of DEMs
Photoclinometry DEM Improved DEM
10
Error Assessment ATM DEM Elevations
ATM elevations - photoclinometry DEM Dh -4.63
m /- 45.06 m
ATM elevations new DEM Dh -0.12 m /- 31.84 m
11
NEW AND IMPROVED SHADED RELIEF MAP
12
UPDATE ON ENVISAT and TerraSAR-X ACQUISTIONS
13
Update on ENVISAT GREENLAND Acquisition
ENVISAT ASAR is C-band with 27 meter resolution
and a 35 day repeat cycle Target areas were the
Peterman, Jakobshavn, and Kangerlussuaq Glaciers
with a 2 cycle repeat coverage for both ascending
and descending geometries. Of the 44 swaths
required to cover the AOI, 39 have been acquired
(5 ascending swaths over Jakobshavn had conflicts
with commercial users) and all but 7 have been
received at OSU.
14
Update on ENVISAT GREENLAND Acquisition
ENVISAT Level 0 (stripline) data was processed
through the VEXEL FOCUS processor to produce RAMS
compatible input data (low resolution quicklook
tifs and associated metadata). This data was
ingested into RAMS software, framed, and checked
for Doppler Ambiguities. The framing information
was then applied to the Level 0 data via the
FOCUS processor for the creation of SLCs (single
look complex). The SLCs were ingested into RAMS
where a binary image product was created and
along track registration preformed.
Interferometry was then run on the framed
pairs. Once the interferometry has been Q/As
velocities are computed.
15
Geo-reference Quicklook Images
16
Initial Envisat Velocity Vectors
17
TerraSAR-X Requested Coverage
18
Velocities from 3-Year Feature Retracking