increasing unmodeled dynamical effects. Roll and pitch biases
are modeled and corrected before obtaining the final
geolocation information. Roll and pitch biases can each be
established because Shuttle attitude changes in roll and pitch
are significantly out of phase (Luthcke et al., 1999).
A first order approach in recovering pointing and range biases
is done using a direct altimetry range residual analysis,
combining spacecraft attitude information with ocean range
residuals (Luthcke et al. 1999). The direct altimetry ocean data
are compared with OSU (Ohio State University) 1995 Mean
Sea Surface Model (Yi, 1995), plus the effects of tides from the
Ray Ocean Altimeter Pathfinder Tide Model (an extension of
the Schrama-Ray Tide Model, 1994), giving a height error for
each laser pulse yielding an ocean surface return. Surface
elevations of the open ocean are known, through measurements
and modeling, to the 12 cm (1 sigma) level, providing a global
reference surface to compare to the altimeter range
measurements throughout the mission. SLA pointing
corrections to the a priori roll and pitch were computed for each
of the SLA-02 observation periods, iterating to solve for
constant biases that reduce the ocean residuals until
convergence was reached. This approach does not include the
smaller contributions to ocean surface height variation from
barotropic pressure («10 cm), earth tides («20 cm) and the time
dependent part of dynamic sea surface topography («50 cm).
SLA-ocean surface height differences on the order of 30 meters
were typically observed before pointing bias estimation, with
larger residuals present in some of the arcs where the Shuttle
attitude exhibited a significant number of maneuvers. After
establishing the attitude corrections, the bounce point
geolocation was recomputed as described above. Average
values for the constant attitude biases obtained from the first
four observation periods (periods without significant
maneuvers) were applied to the data from observations when
attitude biases were difficult to extract from the residuals. Data
from observation period 17 were geolocated applying no bias
corrections for roll and pitch. Figure 3 shows a histogram of
the final SLA-02 ocean surface residuals for all observation
periods analyzed. The mean and standard deviation of ocean
surface residuals is larger for SLA-02 than —01 (Garvin et al.,
1998), possibly due to shorter arcs and greater changes in the
attitude profiles, which could result in unmodeled time-varying
pointing biases during the individual observation periods. In
addition, the more frequent attitude and orbit maneuvers during
SLA-02 observations may have contributed to increased Shuttle
body flexure and non-constant IMU misalignment effects that
have not yet been compensated, and made more difficult to de-
couple orbit and pointing errors. A much smaller contribution
could be attributed to sea surface wave structure, barotropic
pressure and solid Earth tides. We are currently in the process
of working towards better modeling and recovering these
pointing biases and improving the orbits to enhance the
geolocation of selected SLA-02 arcs. Leveling Correction and
Computing Orthometric Elevations.
International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 3W14, La Jolla, CA, 9-11 Nov. 1999
MSS95+Tides -SLA-Ht.
SLA-02 OBS.:1,2,3,4,4a,7,8,9,10,11,12,13,15,16,17,18,19,20 (SDP V2)
% # OBS.
-30 -25 -20 -15 -10 5 0 5 10 15 20 25. .30
METERS
Fig. 3. Histogram of SLA-02 elevation differences
with respect to T/P=based Mean Sea Surface
ocean topography corrected for ocean tides
À leveling correction is applied to the resulting elevation data
for each laser bounce point to correct the effects of long
wavelength orbit errors. The ocean range residuals time series
is smoothed by a sliding boxcar filter with a window length of
120 seconds. The minimum number of ocean surface laser
returns allowed within the window was 50, and a 3 sigma
editing of outlier residuals was performed. The resulting ocean
leveling correction is extrapolated across land areas using a
linear fit to ocean results prior to and after the land. The
leveling correction, provided in the sla02.bp.surface 3
parameter, is a measure of the elevation error that is primarily
due to long-wavelength orbit errors. The geolocation process
yields elevations referenced to the T/P ellipsoid. Orthometric
elevations were computed by subtracting the geoid height at
each laser bounce-point defined by the Earth Geoid Model 96
(EGM96) (Lemoine et al., 1998). This level of processing
constitutes the SLA-02 Standard Data Product Version 2 (SDP
v2).
3. ADDITIONAL PROCESSING
3.1. Return Type Classification
Each laser shot was classified according to a scheme that
distinguishes returns from ocean or land surfaces based on
masks derived from the TerrainBase 5 minute (10 km)
resolution global terrain model (Figure 4). Ocean and land
shots were each further classified as valid returns, from the
Earth surface or clouds, or as non-valid returns, due either to a
range return from background noise or a no-range return (no
backscatter signal detected above the range acquisition
threshold). Ocean surface returns were defined as those whose
orthometric elevations did not depart from sea level (elevation
= 0) by more than 20 meters. Land surface returns had
orthometric elevations within 500 meters of TerrainBase. This
larger land elevation threshold was chosen to account for
inaccuracies in TerrainBase and geolocation errors causing
large elevation discrepancies in high-relief errors. This method
probably overestimates the percentage of land surface returns,
classifying returns from some low altitude clouds as being from
Internatio
the surface. Return
than 500 m and
reference surfaces
(considered to be
classified as noise
TerrainBase and «
above 10,000 m.
classification categ:
shots that have beer
587,172 over-
1,511,828 over-
600,000
550,000
500,000
450,000
400,000
350,000
300,000
250,000
# of Occurrences
200,000
150,000
100,000
50,00€
i
The proportion of
comparable for lai
cloud returns is sig
the ocean, indicati
land areas samp
parameters are als
ISESCP land ©
Vegetation Index (
bounce-point geolc
32. Orientatio
To allow assessm
spreading of the b
vector with respec
altimetry data wer
coordinates and
