Professional Documents
Culture Documents
Dem Control in Arctic Alaska With Icesat Laser Altimetry
Dem Control in Arctic Alaska With Icesat Laser Altimetry
Abstract—Use of Ice, Cloud, and land Elevation Satellite Optical photogrammetry and interferometric synthetic aperture
(ICESat) laser altimetry is demonstrated for control of a digital radar (InSAR) both depend on GCPs to remove translational
elevation model (DEM) that is synthesized from repeat-pass ERS-1 and scale errors, and to adjust the elevations of digital elevation
and 2 synthetic aperture radar (SAR) imagery using interferomet-
ric SAR (InSAR). Our study area is 15 650 km2 of the Barrow, models (DEMs) derived from multiple photographs or images
AK coastal plain adjacent to the Arctic Ocean; a vast expanse of of the region of interest. Acquisition of GCPs usually involves
tundra, lakes, and arctic wetlands of such low relief as to be nearly positioning markers identifiable in spaceborne or airborne im-
devoid of terrain features. The accuracy of the ICESat-derived agery on the ground prior to acquisition of the imagery.
elevation measurements is assessed by comparison with dif- In most instances, GCP acquisition requires a significant
ferential global positioning system (DGPS) data acquired
along ICESat ground tracks. The ICESat-derived elevations logistical effort to ensure a broad distribution of points, as
have a mean accuracy, relative to the DGPS elevations, of well as expensive global positioning system (GPS) equipment
−0.01 ± 0.18 m. ICESat-derived elevations on the Arctic coastal and sophisticated postprocessing of the data. In populated re-
plain provide an excellent source for DEM control. We employ gions with road networks, GCPs are easy to acquire. However,
the ICESat-derived ground control points (GCPs) in two distinct in remote regions, GCP acquisition typically involves costly
InSAR processing steps: 1) selected points are used to perform
baseline refinements, which improves the ERS-1 and 2 interfero- deployments via helicopter or light airplane. Thus, there is
grams and 2) the ICESat-derived GCP position data (latitude, lon- a strong interest in new sources of GCPs that can be used
gitude, elevation) are then used as control in mosaicking multiple in conjunction with spaceborne or airborne remote sensing
InSAR-derived DEMs. The resulting ICESat-controlled DEM has methods for production of accurate DEMs.
a mean accuracy of −1.11 ± 6.3 m relative to an independent In January 2003, National Aeronautics and Space Admin-
standard, which is a commercial airborne InSAR-derived DEM
having 0.5 m rms accuracy. This easily meets DTED-2 standards istration (NASA) launched the Geoscience Laser Altimeter
and suggests that DEMs derived using only ICESat altimetry for System (GLAS) into high polar orbit onboard the Ice, Cloud,
ground control would meet similar standards in other regions of and Land Elevation Satellite (ICESat). A major objective of this
low relief. spaceborne laser altimeter system, with orbital coverage from
Index Terms—Global positioning system, interferometry, ter- latitude 86◦ N to 86◦ S, is to provide elevation measurements
rain mapping, satellite applications, synthetic aperture (RADAR). of the Earth’s topography with accuracy sufficient to serve as a
terrestrial reference frame [1].
I. I NTRODUCTION We investigate the use of ICESat-derived elevations and their
associated positional data as an alternative source of ground
(ERS)-1 and 2 Tandem Mission, which flew during fall control that is well distributed is an essential requirement for
1995 to spring 1996, for DEM generation. mapping remote regions such as Alaska [5].
5) A high-resolution commercial airborne InSAR-derived Despite manufacturing problems with the onboard laser
DEM is available for the Barrow region that can be diodes, there have been nine 33+-day periods of observations.
employed as an independent standard of accuracy. Throughout this time period, the ICESat Science Team has
6) Conventional GCPs measured with differential GPS continued to refine and implement processing algorithms for the
(DGPS), although not widely distributed, are also avail- GLAS data. The final data set used in this study was acquired
able in the Barrow region for independent assessments of from the National Snow and Ice Data Center, as of May 2006.
accuracy.
We first describe a field campaign for acquisition of fast-
static and kinematic GPS (KGPS) data along ICESat ground III. P ROJECT D ESIGN
tracks within approximately 50 km of Barrow, carried out The intent of our project is to independently evaluate the ac-
during March 2004, and the use of these data to verify the curacy and repeatability of the ICESat data and to demonstrate
accuracy of ICESat altimetry in this region. We then demon- its suitability for use as ground control in the generation of a
strate a method for incorporating ICESat altimetry for control DEM using spaceborne repeat-pass InSAR. The Barrow region
in the generation of an InSAR-derived DEM of the Barrow was selected as our study site due to its low topography relief,
region. The accuracy of this DEM is then determined by abundance of lakes, and availability of high-quality scientific
comparison with an independent standard, and the results are data for the region. Our approach follows two paths.
summarized. 1) Use data from a field campaign to evaluate the accuracy
and precision of the ICESat-derived elevations on lakes
II. P OTENTIAL OF ICESat A LTIMETRY in the study area. This is achieved by comparing ICESat
FOR DEM C ONTROL elevations with those of precision DGPS. The choice of
lakes ensures flat topography that will be immune to
ICESat, which was launched in January 2003, is part
offset errors in the locations of the laser footprints. In
of NASA’s Earth Observing System. The GLAS onboard
addition, the flat surface ensures a simple Gaussian return
ICESat measures ice sheet elevations, changes in elevation
waveform from GLAS, making it easier to interpret the
over time, height profiles of clouds and aerosols, land eleva-
centroids as average footprint elevations. We acquire the
tions and land cover, and sea ice thickness. Because of its
DGPS data using both kinematic measurements and long
global extent and unprecedented vertical accuracy, ICESat was
residence, surveying at multiple points along the ICESat
considered as a possible means for acquiring ground control
ground tracks.
for accurate DEMs of the Earth’s topography. Specifically,
2) Utilize the ICESat data as ground control for the de-
ICESat was intended to be used “as a reference framework
velopment of an InSAR-derived DEM of the Barrow
to evaluate and improve the accuracy of topographic maps
Peninsula. The utility of the data is to be determined
acquired by other airborne and space-based methods such as
for each step that requires ground control, specifically,
conventional stereo-photogrammetry and radar interferometry”
baseline determination and mosaic processing.
(NASA, http://ICESat.gsfc.nasa.gov).
The GLAS instrument is an altimeter that continuously emits Three data sets were used in this study: 1) DGPS for analysis
laser pulses at a rate of 40 Hz. Round trip times, coupled with of GLAS data quality at specific locations along the ICESat
precise knowledge of satellite location and orientation (from ground path; 2) ERS-1 and 2 SAR image pairs acquired with
onboard GPS, a star tracker, and ground laser tracking), permit one-day time separation during the fall 1995 to spring 1996
computation of elevation and ground position for each measure- Tandem Mission, for assessment of the utility of ICESat ground
ment made. As ICESat orbits, GLAS receives backscattered control for DEM generation; and 3) a high-resolution commer-
waveforms from a succession of ∼70-m footprints with along- cial DEM derived from airborne SAR data, for assessment of
track separation of 172 m. Precise footprint positions are de- the accuracy of our InSAR-derived ICESat-controlled DEM.
rived from the 1064-nm channel, while dual 532- and 1064-nm
channels provide information on clouds and aerosols [2].
IV. F IELD C AMPAIGN
ICESat flies in a near-polar orbit with a 91-day exact-repeat
cycle and a 33-day near-repeat subcycle. Although the ground- In March 2004, a GPS survey was performed in the Barrow
track spacing is relatively large at equatorial and temperate region, using two survey-grade GPS receivers with ground-
latitudes, track separation in the Arctic is much less, on the plane equipped antennas: one as a remote base station for
order of tens of kilometers, because of the orbital convergence individual ponds and lakes, and the other as a rover unit for
at the Poles. The elevation accuracy of GLAS, of course, fast-static and kinematic measurements. Two continuous base
depends on terrain slopes, vegetation, and snow cover. At the stations (SG27 and BASC) also operated in Barrow. We also
time of this report, ICESat-derived elevation accuracy (bias) surveyed a benchmark (S123) in Barrow with a third survey-
has been shown to be < 2 cm with a precision of < 3 cm grade receiver, which remained on-site during the field cam-
[3], [4]. This precision and accuracy, although not as high as paign. The Barrow Arctic Science Consortium (BASC) base
that obtainable from DGPS, appears to be more than sufficient station and site S123 did not operate on all days of the field cam-
for control of DEMs. The availability of high-quality geodetic paign, but data from SG27 provided continues operation for all
3712 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 11, NOVEMBER 2007
Fig. 1. SAR amplitude imagery of the Barrow Peninsula. Overlays indicate lakes where GPS data were acquired and the ICESat ground tracks. Kinematic survey
tracks are shown as well. Sea-ice-covered Arctic Ocean is at upper right and left. SAR image, ©ESA 2003.
days. The GPS measurements were timed to be approximately lake. The fast-static surveys also used tripod-mounted antennas.
coincident with the February–March 2004 ICESat acquisitions. All GPS receivers sampled data at an interval of 30 s. The fast-
Field support was provided by the BASC. It provided logis- static rover measurements yielded important data on possible
tical support in the form of transportation, accommodations, surface variability across the lake, whereas the associated snow
snow machines, sleds, field equipment, and trained personnel. measurements yielded information on the thickness and vari-
Of the seven days spent in Barrow (March 19–26, 2004), ability of the snow cover.
whiteout and high wind-speed conditions curtailed field oper- In all instances, we attempted to reference the GPS measure-
ations on three days, leaving four days (21, 22, 24, 26) for ments to the ice surfaces of the lakes. This technique probably
measurements. Even the “good” days presented hardships in provided accurate measurements in most instances. At some
the form of extreme cold, i.e., temperatures with wind chill as sites, however, we found that stratified snow had hardened
low as minus 50 ◦ C (minus 58 ◦ F). Over the broad geographic into ice, and this may have contributed to measurement error
area of the study area, travel via snow machine was slowed by by elevating the GPS elevations several centimeters above the
sastrugi (snow dunes), whereas the harsh pounding of the sleds actual surfaces of the lake ice.
caused equipment damage. Since the laser returns seen by ICESat come from the top
Static GPS data were acquired on frozen lakes because the surface of the snow, it is important to ascertain the mean snow
level lake surfaces were expected to minimize the positional thicknesses on a scale consistent with the footprints of ICESat.
errors in the ICESat footprints. Had measurements been made To achieve this, we measured the snow thickness at the site of
on land, any x/y disparities between the ICESat footprints and each fast-static GPS measurement, then made a series of ten
GPS measurements may have led to errors caused by the mild snow thickness measurements, spaced every 5 m, in each of
topography. Fig. 1 is a SAR image with overlays that depict the the four cardinal directions. These measurements were later
sites of the ground survey as well as the ICESat tracks used in averaged to provide a mean snow thickness in the region of
the analysis. the corresponding fast-static GPS measurement. We biased the
A total of 21 GPS rover measurements were carried out on GPS measurements by this mean snow thickness to estimate
a total of five lakes during the field study. At each survey site, the elevation of the snow surface as seen by ICESat at that
we set up a local base station on a tripod near the center of location. It should be noted that since the GPS survey and the
the selected lake. We then conducted a series of up to five ICESat passes were not coincident in time, the possibility exists
20-min fast-static surveys, cardinally distributed across the for an additional bias being introduced by snowfall between
ATWOOD et al.: DEM CONTROL IN ARCTIC ALASKA WITH ICESAT LASER ALTIMETRY 3713
the ICESat passes and the survey. This concern is addressed The final analysis step was to adjust all regional and local
in Section V. solutions together to generate a single self-consistent set of co-
In addition, KGPS data were acquired along the ICESat ordinates. This final step is necessary because we used multiple
ground tracks. One of the KGPS acquisitions coincided with base stations in Barrow and did not fix any coordinates a priori.
the ICESat ground tracks from November 14, 2003 (0349) Each lake was surveyed on only one day, but the sites in Barrow
and February 21, 2004 (1331), whereas the second coincided were common to all days, as were the 66 continuous GPS
with an ICESat ground track of March 3, 2004 (0149). During sites included in the solutions. Antenna heights, as recorded
the two kinematic surveys, the GPS receiver was mounted on in the field, were subtracted from the results to yield lake ice
a snow machine, which was slowly and consistently driven elevation.
along a presumed ground path. The liquid crystal display of the Postprocessing of the KGPS measurements was performed
GPS receiver could not sustain the cold for extended periods, using the Trimble Office software package with same-day
so waypoints were chosen to enable the driver to stay on acquisition data files from six GPS stations of the National
course. Oceanic and Atmospheric Administration Continuously Oper-
ating Reference Station (NOAA CORS) network in Alaska.
The NOAA CORS, along with the two local base stations, pro-
V. ICESat V ERSUS DGPS: A CCURACY AND P RECISION
vided a network of eight stations for derivation of differential-
To compare elevations from ICESat to elevations measured corrected least-squares solutions to the KGPS positions along
in the field with GPS, we must take care to determine heights three profiles crossing tundra and frozen lakes. The height of
from both techniques in the same coordinate system. ICESat the antenna mounted on the snow machine was subtracted to
measures ellipsoid heights relative to the TOPEX/Poseidon yield snow surface elevations.
ellipsoid data. A program was employed to convert the ellipsoid Our data analysis focused on near-concurrent comparisons
heights from TOPEX/Poseidon to the WGS-84 ellipsoid data. of ICESat-derived elevations and DGPS lake measurements. Of
We computed heights from GPS in the ITRF97 reference frame, the five lakes surveyed, two lakes were eliminated because there
and expressed them as ellipsoid heights above the WGS-84 are no ICESat data at the time of the survey. The February 21,
ellipsoid. The present definition of WGS-84 coordinates is 2004 ICESat pass missed the small lake, U1, and there were
based on the International Terrestrial Reference System (ITRS); no 2004 passes associated with U3 due to a change in ICESat
so at DEM accuracy levels, ITRF and WGS-84 are equiva- scheduling. Thus, the analysis focused on Lakes G8, G3 and
lent [8]. Because we have a mixture of long and short static U5, using ICESat data from the February 21 and March 3, 2004
occupations, fast-static surveys, and kinematic surveys, we ICESat passes.
used a nested analysis strategy to place all results in ITRF97 Table I summarizes the results of the field survey and the as-
while minimizing errors due to tropospheric delays and other sociated ICESat-derived elevations. All vertical measurements
potential biases. Static and fast-static GPS data from the field are ellipsoid-based in the WGS-84 data. Lake G3 was surveyed
campaign were analyzed using the GIPSY-OASIS II software at one remote site and five fast-static sites and compared with
developed at Jet Propulsion Laboratory (JPL) [6], driven by ICESat passes 1331 (February 21, 2004) and 0149 (March 3,
in-house written analysis scripts. The GIPSY software uses 2004). Lake G8 was surveyed at one remote site and two fast-
undifferenced phase and pseudorange measurements, and offers static sites and compared with ICESat pass 1331 (February 21,
a great deal of flexibility in models for estimated parameters. 2004). Lake U5 was surveyed at one remote site and one fast-
For the static and fast-static surveys, we decimated data from static site and compared with ICESat pass 0149 (March 3,
all sites to an epoch every 5 min. 2004). The table reflects individual fast-static results as well
First, we included data from the three sites in Barrow (SG27, as mean and standard deviation values for the sum of all lakes.
BASC, and S123) in a regional solution, along with data The GPS survey elevations were corrected by the snow
from a network of 66 continuously operating GPS stations thickness for comparison with the ICESat-derived mean lake
distributed across the Northern Hemisphere, with the strategy elevations. To determine whether snowfall accumulation be-
of Freymueller et al. [7]. In this regional solution, we esti- tween the ICESat passes and GPS survey contributed additional
mated time-dependent tropospheric delays at all sites, plus an biases, we investigated the National Weather Service (NWS)
azimuthal-dependent tropospheric gradient. No sites were fixed climatological records for Barrow. At the time of the February
in this solution, and the JPL nonfiducial (frame-free) orbits and March ICESat passes, the measured snow depth in Barrow
were fixed. The regional solution, based on 24-hr surveys, is was 9 in. By the time of our field campaign, the depth had
then aligned with ITRF97 with subcentimeter precision using a increased to 11 in on March 21 and 12 in on March 25. This ad-
seven-parameter transformation. Second, we estimated a local ditional accumulation was incorporated into the snow thickness
solution using only the Barrow sites and lake sites (lake base correction in Table I. Comparison of the corrected DGPS and
stations and fast-static sites), and resolving phase ambiguities. ICESat measurements yields a mean difference of 1 cm, well
In this local solution, we estimated only a constant tropospheric within the standard deviation of the individual measurements.
delay for each site, which amounts to an assumption that the Fig. 2 shows the results of one kinematic survey, along
time variation in the troposphere is constant for all sites (the with coincident ICESat tracks from ICESat passes 1331
estimated zenith tropospheric delays differed by no more than (February 21, 2004) and 0349 (November 14, 2003). The snow-
1–2 cm from site to site). No sites were fixed in this solution machine traverse approximated the corresponding ICESat
either, and the JPL orbits in ITRF97 were fixed. ground track. These measurements enabled us to assess ICESat
3714 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 11, NOVEMBER 2007
TABLE I TABLE II
DGPS AND SNOW THICKNESS MEASUREMENTS PERFORMED ON LAKES KGPS ELEVATIONS ACQUIRED MARCH 22, 2004 COMPARED WITH
G3, G8, AND U5 ON MARCH 19–26, 2004 COMPARED WITH MEAN ICESat ELEVATIONS ACQUIRED NOVEMBER 14, 2003,
ICESat MEASUREMENTS ON FEBRUARY 21, 2004 (1331) AND MARCH 3, FEBRUARY 21, 2004, AND MARCH 3, 2004. CORRECTION FOR SNOW
2004 (0149). MEAN SNOW THICKNESS WAS ADDED TO DGPS ACCUMULATION BETWEEN ICESat PASSES AND SURVEY DATE
ELEVATION ON EACH LAKE TO PERMIT A SNOW SURFACE REFERENCED IS BASED ON DATA FROM NWS OFFICE IN BARROW. ALL
COMPARISON. CORRECTION FOR SNOW ACCUMULATION BETWEEN HEIGHTS ARE RELATIVE TO WGS-84 ELLIPSOID
ICESat PASSES AND SURVEY DATES ARE BASED ON DATA
FROM NWS OFFICE IN BARROW
TABLE III
TABLE SHOWING THE ORBITS THAT WERE USED IN TESTING
THE A PPLICATION OF ICESat D ATA AS G ROUND
CONTROL FOR InSAR PROCESSING
Fig. 4. (a) Unwrapped phase using perpendicular baseline derived from resti-
tuted state vectors, showing that a phase ramp dominates all terrain features.
(b) Use of iteratively culled ICESat GCPs distributed along the superposed orbit
ground tracks for baseline refinement has resulted in removal of the phase ramp.
Terrain features, mostly tundra lakes, are now visible.
Fig. 6. Top row, (a)–(c): Color-ramped individual ICESat-controlled DEMs. Bottom row, (a)–(c): Error distributions for the three individual ICESat-
controlled DEMs.
Fig. 7. (a) Color-ramped final DEM mosaic. (b) Color-ramped Star-3i reference DEM. (c) Error distribution for final ICESat-controlled DEM mosaic (a)
compared with Star-3i DEM (b).
in Fig. 6 shows a color-ramped DEM for each of the three height errors are significant for the three individual DEMs,
ERS SAR image pairs. The bottom row shows the elevation but these errors are minimized in the final mosaic due to:
errors relative to the Star-3i standard. Fig. 7 compares the 1) averaging of the three independent DEM products, which
spaceborne InSAR-derived ICESat-controlled composite DEM reduced the random errors and 2) the use of ICESat heights for
with the Star-3i airborne-derived DEM and depicts the resulting vertical control. Upon completion of the second pass mosaic,
error distribution. the offset of the ICESat-controlled DEM relative to the Star-3i
Table IV summarizes the relative accuracies of our three DEM is −1.11 ± 6.30 m (Star-3i lower), as graphically shown
individual InSAR DEMs and our final DEM mosaic. The in Fig. 8.
3718 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 11, NOVEMBER 2007
TABLE IV to the next, the error maps exhibit spatial variability throughout
MEAN ELEVATION DIFFERENCES BETWEEN ICESat-CONTROLLED DEMs
AND S TAR -3i InSAR-D ERIVED DEM OF THE S AME R EGION .
the region of study. Inadequate baseline refinement would lead
TABLE INCLUDES ERROR ANALYSES FOR INDIVIDUAL InSAR DEMs to an error ramp across any individual DEM, and no such ramp
(FIRST THREE ROWS) AS WELL AS THE FINAL exists in any of the three error maps in Fig. 6(a)–(c) (lower row).
DEM MOSAIC (LAST ROW)
Thus, we can rule out InSAR processing errors as the cause.
The most likely explanation for the spatially varying resid-
ual errors is tropospheric water vapor [16]–[18]. Tropospheric
water alters the dielectric constant of the atmosphere through
which the microwaves are transmitted and, therefore, changes
the accumulated phase. Changes in tropospheric water vapor
between repeat passes can lead to regional changes in the
interferogram over distances of tens of kilometers, consistent
with our DEM elevation errors.
The spatially varying and significant role of water vapor on
our DEM accuracy is not surprising. Point Barrow is the most
northerly point in Alaska, projecting into the Arctic Ocean. The
ERS-1 and 2 SAR images used in our study were all acquired
in winter, when the coastal waters were largely covered with
sea ice. Changing ice dynamics along the coast provides a
mechanism for marine moisture to enter the atmosphere in
a variable manner. That is, spatially variable offshore leads
release sizeable quantities of water vapor into the atmosphere.
This temporal and spatial variability is a mechanism capable
of altering the tropospheric water vapor near the coast over the
one-day repeat cycle of the ERS satellites. In turn, this may
cause the observed systematic elevation errors. Further work
will be required to establish this explanation.
DEM and the ICESat elevations based on an assumption that [3] H. A. Fricker, A. Borsa, B. Minster, C. Carabajal, K. Quinn, and
the latitude–longitude locations of the ICESat footprints are B. Bills, “Assessment of ICESat performance at the salar de Uyuni,
Bolivia,” Geophys. Res. Lett., vol. 32, no. 21, pp. L21 806.1–L21 806.5,
accurate. In this study area, however, lack of coregistration 2005. DOI:10.1029/2005GL023423.
should not present a significant problem. The reasons are as [4] C. C. Carabajal and D. J. Harding, “SRTM C-band and ICESat laser
follows: 1) the 10-m horizontal positional accuracy of the altimetry elevation comparisons as a function of tree cover and relief,”
Photogramm. Eng. Remote Sens., vol. 72, no. 3, pp. 287–298, 2006.
ICESat data is exceptional and 2) the positional accuracy of the [5] J. Sauber, B. Molnia, C. Carabajal, S. Luthcke, and R. Muskett,
spaceborne ERS-1 and 2 SAR data is better than 100 m, based “Ice elevations and surface change on the Malaspina Glacier, Alaska,”
on a comparison of ICESat elevation profiles and the observed Geophys. Res. Lett., vol. 32, no. 23, pp. L23 S01.1–L23 S01.4, 2005.
DOI:10.1029/2005GL023943.
locations of lakes in the SAR imagery. That is, even with an [6] J. F. Zumberge, M. B. Heflin, D. C. Jefferson, M. M. Watkins, and
absence of image-recognizable GCP locations, the positions of F. H. Webb, “Precise point positioning for the efficient and robust analysis
the ICESat-derived GCPs are known to within 10 m, and the of GPS data from large networks,” J. Geophys. Res., vol. 102, no. B3,
corresponding DEM locations, prior to use of the ICESat data pp. 5005–5018, 1997.
[7] J. T. Freymueller, S. C. Cohen, and H. J. Fletcher, “Spatial variations
for control, were generally within about 100 m. in present-day deformation, Kenai Peninsula, Alaska, and their implica-
Given the demonstrated accuracy of the ICESat data, the an- tions,” J. Geophys. Res., vol. 105, no. B4, pp. 8079–8102, 2000.
swer to the question of whether these data can be integrated into [8] National Imagery and Mapping Agency, Department of Defense World
Geodetic System 1984, Jan. 3, 2000.
the processing of a DEM is “yes.” A key concern was whether [9] G. Sun, K. J. Ranson, J. Bufton, and M. Roth, “Requirement of ground
the sparse geographic distribution of the ICESat ground tracks tie points for InSAR DEM generation,” Photogramm. Eng. Remote Sens.,
is adequate for DEM control. On the Barrow Peninsula, the vol. 66, no. 1, pp. 81–85, 2000.
[10] H. Zebker and R. Goldstein, “Topographic mapping from interferometric
ICESat orbits are sufficiently closely spaced for this purpose, SAR observations,” J. Geophys. Res., vol. 91, no. B5, pp. 4993–4999,
with a separation on the order of tens of kilometers. 1986.
We employed ICESat data in two key steps: 1) refinement [11] H. A. Zebker, C. L. Werner, P. Rosen, and S. Hensley, “Accuracy of topo-
graphic maps derived from ERS-1 interferometric radar,” IEEE Trans.
of the interferometric baselines and 2) as ground control for Geosci. Remote Sens., vol. 32, no. 4, pp. 823–836, Jul. 1994.
the mosaicking of three individual spaceborne InSAR-derived [12] O. Lawlor, T. Logan, R. Guritz, R. Fatland, S. Li, Z. Wang, and
DEMs. The ICESat-derived elevations admirably serve in each C. Olmsted, “Generation of fine resolution DEM at test areas in Alaska
using ERS tandem pairs and precise orbital data,” in Proc. 12th Int. Conf.
of these production steps because the convergence of the Appl. Geologic Remote Sens., Denver, CO, Nov. 17–19, 1997.
ICESat ground tracks in the high Arctic yields a sufficient [13] D. R. Fatland and C. S. Lingle, “Analysis of the 1993–95 Bering Glacier
density of points within the 100-km swaths of the SAR. This (Alaska) surge using differential SAR interferometry,” J. Glaciol., vol. 44,
study clearly shows that one benefit of the ICESat mission will no. 148, pp. 532–546, 1998.
[14] R. Guritz, M. Ayers, T. Logan, S. Li, and O. Lawlor, “Automated DEM
be to provide an extensive data set of control points to support production using ESA tandem mission data for the Caribou-Poker Creek
future mapping of the polar regions. LTER Watershed, Alaska,” in Proc. Fringe, Leige, Belgium, Nov. 10–12,
1999.
[15] S. H. Lee, S. Baek, and C. K. Shum, “Multi-temporal, multi-resolution
data fusion for Antarctica DEM determination using InSAR and
ACKNOWLEDGMENT altimetry,” in Proc. IGARSS, 2005, vol. 4, pp. 2827–2829.
[16] R. F. Hanssen, T. M. Weckwerth, H. A. Zebker, and R. Klees, “High-
The authors would like to thank Dr. P. Salamonowicz of resolution water vapor mapping from interferometric radar measure-
NGA for traveling to the University of Alaska Fairbanks for ments,” Science, vol. 283, no. 5406, pp. 1297–1299, 1999.
initial discussions of this paper; NASA’s SENH and GSFC’s [17] Z. Li, J. P. Muller, P. Cross, and E. J. Fielding, “Interferometric synthetic
aperture radar (InSAR) atmospheric correction: GPS, Moderate Reso-
IRAD programs for support of J. M. Sauber, R. Gens, and lution Imaging Spectroradiometer (MODIS), and InSAR integration,”
B. Kerin of the Alaska Satellite Facility and O. Lawlor of the J. Geophys. Res., vol. 110, no. B3, pp. B03 410.1–B03 410.10, 2005.
UAF Department of Mathematics and Statistics for assistance DOI:10.1029/2004JB003446.
[18] D. Moisseev, R. Hanssen, and J. Sabater, “Towards an atmosphere free
with programming and InSAR processing; A. Kaufman of interferogram; first comparison between ENVISAT’s ASAR and MERIS
the Geophysical Institute/UAF for GPS equipment assistance; water vapor observations,” in Proc. IGARSS, 2003, vol. 5, pp. 2977–2980.
M. Jeffries of the Geophysical Institute/UAF and C. Duguay of
the University of Waterloo for advice and equipment that we
used for snow depth measurement on the Barrow Peninsula;
the staff of the Barrow Arctic Science Consortium for logis-
tical support; P. Anashoak for invaluable local knowledge and
support during our Barrow field campaign; J. List, C. Markon,
H. C. Brown, and D. B. Kintz of the USGS, as well as Donald K. Atwood received the Ph.D. degree in
G. Panos of the BLM, for providing Star-3i DEMs; W. Manley physics from the Massachusetts Institute of Technol-
ogy (MIT), Cambridge, in 1982.
of the University of Colorado for access to geodetic control He was with Bell Laboratories and Raytheon’s
data; and E. Chapin for consultation regarding use of the JPL Research Division in the field of microlithography
Multi-mosaic software. and with MIT Sea Grant in underwater acoustics.
He was with Raytheon, first as a Technical Advisor
for international environmental programs, then as
a Manager in atmospheric remote sensing with the
R EFERENCES Goddard Space Flight Center. In 2000, he became the
[1] B. E. Schutz, H. J. Zwally, C. A. Shuman, D. Hancock, and J. P. DiMarzio, Director of Science Support for the U.S. Antarctic
“Overview of the ICESat mission,” Geophys. Res. Lett., vol. 32, no. 21, Program. He is currently the Manager of the Remote Sensing Support Center,
pp. L21 801.1–L21 801.4, 2005. DOI:10.1029/2005GL024009. Alaska Satellite Facility/Geophysical Institute, University of Alaska, Fairbanks,
[2] H. J. Zwally et al., “ICESat’s laser measurements of polar ice, atmosphere, where he is responsible for supporting the U.S. SAR community and contribut-
ocean, and land,” J. Geodyn., vol. 34, no. 3, pp. 405–445, Oct. 2002. ing to special projects in remote sensing.
3720 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 11, NOVEMBER 2007
Richard M. Guritz has been a Special Projects Jeanne M. Sauber received the Ph.D. degree in
Manager with the Alaska Satellite Facility geophysics from the Massachusetts Institute of Tech-
(ASF)/Geophysical Institute, University of Alaska, nology, Cambridge, in 1988.
Fairbanks since 2000. He specializes in inter- She is currently a Geophysicist with the Planetary
ferometric applications of SAR data in projects that Geodynamics Laboratory, NASA Goddard Space
have included a NASA Alaska DEM project and an Flight Center, Greenbelt, MD. The NASA research
NGA north of 60 feasibility study; the latter was to she has led include studies to measure and model
ascertain the feasibility of using ERS-1/2 Tandem ongoing crustal deformation across south central
Mission SAR data to extend to the high Arctic the Alaska using GPS, analysis of topography and sur-
topographic mapping carried out with SAR data face change data to quantify the interaction of
from the Shuttle Radar Topography Mission which glaciers and tectonics, and to use SRTM and GPS
flew in February 2000. Prior to 2000, he managed the Advanced Product data to model the earthquake seismic cycle and constrain the mechanics
Development Group at ASF and was responsible for SAR user tools support of shallow subduction zones. From 1999 to 2006, she was on the ICESat
and SAR algorithm and prototype system development. He has 15 years of Calibration and Validation Team at Goddard.
experience working with SAR data.
Mr. Guritz is the recipient of a NASA Group Achievement Award (for
contributions to the Radarsat Antarctic Mapping Program, 2000), a NASA Jeffrey T. Freymueller received the B.Sc. degree
Contractor Excellence Award (1999), a NASA Group Achievement Award (for in geophysics from the California Institute of Tech-
contributions to the Alaska SAR Development Team, 1992), and a Geophysical nology, Pasadena, in 1985 and the M.S. and Ph.D.
Institute Outstanding Employee of the Year Award (1989). degrees from the University of the South, Columbia,
SC, in 1988 and 1991, respectively.
He is currently a Professor of geology and geo-
Reginald R. Muskett received the B.S. degree in ge- physics with the Geophysical Institute and a Faculty
ology from the University of Alabama, Birmingham, Member of the Department of Geology and Geo-
in 1983, the M.S. degree in geology (geophysics) physics, University of Alaska, Fairbanks. He has au-
from the University of Akron, Akron, OH, in 1998, thored numerous research papers on the application
and the Ph.D. degree in geophysics from the Uni- of GPS to global plate tectonics and seismology.
versity of Alaska, Fairbanks, under C. S. Lingle, Dr. Freymueller has been a member of the Steering Committee, University
in 2007. NAVSTAR Consortium (UNAVCO), since 1995 and the Chairman, since
In 1985, he was with the Defense Mapping 1998. He has also served on committees and panels of the Geodesy Section
Agency, Hydrographic and Topographic Center, U.S. of the American Geophysical Union and is currently the Geodesy Section
Department of Defense, where he became a Senior President. He is the recipient of the 1992 NASA Group Achievement Award for
Cartographer in the Digital Products Department in Development of OASIS/GIPSY Global Positioning System Analysis Software.
1992. He is currently with the University of Alaska.
Dr. Muskett is a member of the American Geophysical Union, the European
Geoscience Union, and the International Association of Geodesy. He is the
recipient of many awards for service in support of operations.