done such that the observations v.;th the smallest associated error estimate was
selected. This sel~ted data set consisted of 755 observations.
The observations were also used to calculate along-track deflections of the
vertical. The along-track deflections of the vertical were calculated by simply taking
the difference between consecutive points divided by the distance. This resulted in
very noisy observations, where the observations with deviations larger than 10" or
with a standard de\iation larger than 4" were rejected as outliers. The resulting 2430
observations had a standard deviation of 2.9".
Empirical auto and CIOSScovariances were computed from the
altimeter
height
observations
, .
I
,
re I ati ve I y to the 0 S LT91 A geoid as
~UQ. -7----::-----=-----+------I:::::::",!..,L---+-E~=__'"~~'_7i
wen as from the marine gra\ity data
using the program EMPCOV. The
EMPCOV program computes such
empirical covariance and CIOSScovarians functions from the observed
data and is a part of the GRAVSOFfprogram package (Tscherning et al.
1992, 1994). Similarly, a covariance
function was computed using the
stacked 35 day repeat data As
expected the sta:ked data has a
slightly smaller variance than the
original data (0.28 m:! instead of 0.30
us
a.aa
UI5
Figure 4 Free air gravity anomalies.
OSU91A contribution removed. Units are in
m 2 ) .
The covariance function of the mgal.
observed marine gravity data was fined by an analytic expression using the COVFIT
program (Knudsen, 1987, 1988). COVFIT is also part of the GRA VSOFf -program
package and fits the empirical covariance function using an analytic base model by
Tscheming & Rapp, (1974) like
where Ej(T,T) are the error degree-variances associated with the OSU91A, R is the
mean earth radius, RB is the radius of a Bjerhammar sphere (RB
OJ are the S
From COVHf the depth to the Bjerhammer sphere R-RB is estimated to 7.3 kID
and the error degree variances factor of the OSU91A field to 0.375. The total gravity
variance is 173.0 mgal:!. The empirical and the fined analytic covariance functions can
221
selected. This sel~ted data set consisted of 755 observations.
The observations were also used to calculate along-track deflections of the
vertical. The along-track deflections of the vertical were calculated by simply taking
the difference between consecutive points divided by the distance. This resulted in
very noisy observations, where the observations with deviations larger than 10" or
with a standard de\iation larger than 4" were rejected as outliers. The resulting 2430
observations had a standard deviation of 2.9".
Empirical auto and CIOSScovariances were computed from the
altimeter
height
observations
, .
I
,
re I ati ve I y to the 0 S LT91 A geoid as
~UQ. -7----::-----=-----+------I:::::::",!..,L---+-E~=__'"~~'_7i
wen as from the marine gra\ity data
using the program EMPCOV. The
EMPCOV program computes such
empirical covariance and CIOSScovarians functions from the observed
data and is a part of the GRAVSOFfprogram package (Tscherning et al.
1992, 1994). Similarly, a covariance
function was computed using the
stacked 35 day repeat data As
expected the sta:ked data has a
slightly smaller variance than the
original data (0.28 m:! instead of 0.30
us
a.aa
UI5
Figure 4 Free air gravity anomalies.
OSU91A contribution removed. Units are in
m 2 ) .
The covariance function of the mgal.
observed marine gravity data was fined by an analytic expression using the COVFIT
program (Knudsen, 1987, 1988). COVFIT is also part of the GRA VSOFf -program
package and fits the empirical covariance function using an analytic base model by
Tscheming & Rapp, (1974) like
where Ej(T,T) are the error degree-variances associated with the OSU91A, R is the
mean earth radius, RB is the radius of a Bjerhammar sphere (RB
OJ are the S
From COVHf the depth to the Bjerhammer sphere R-RB is estimated to 7.3 kID
and the error degree variances factor of the OSU91A field to 0.375. The total gravity
variance is 173.0 mgal:!. The empirical and the fined analytic covariance functions can
221