Large Scale Ocean Circulation from the GRACE GGM01 Geoid

Document technical information

Format pdf
Size 256.0 kB
First found Jun 9, 2017

Document content analysis

not defined
no text concepts found





Copyright 2004 American Geophysical Union. Further reproduction or electronic distribution is not permitted.
Large Scale Ocean Circulation from the GRACE GGM01 Geoid
B. D. Tapley, D. P. Chambers, S. Bettadpur, and J. C. Ries
Center for Space Research, The University of Texas at Austin, Austin, Texas,USA
The GRACE Gravity Model 01 (GGM01), computed
from 111 days of GRACE K-band ranging (KBR) data, is
differenced from a global mean sea surface (MSS)
computed from a decade of satellite altimetry to
determine a mean dynamic ocean topography (DOT). As
a test of the GGM01 gravity model, large-scale zonal and
meridional surface geostrophic currents are computed
from the topography and are compared with those derived
from a mean hydrographic surface. Reduction in residual
RMS between the two by 30-60% (and increased
correlation) indicates that the GGM01 geoid represents a
dramatic improvement over older geoid models, which
were developed from multiple satellite tracking data,
altimetry, and surface gravity measurements. For the first
time, all major current systems are clearly observed in the
DOT from space-based measurements.
1. Introduction
An Earth geopotential model that has both high
accuracy and spatial resolution is a requirement for a
number of contemporary studies in geophysics and
oceanography. Wunsch and Gaposchkin [1980] (hereafter
referred to as WG80) discussed how accurate sea surface
height (SSH) measurements from satellite altimetry could
be combined with a precise geoid to compute the absolute
dynamic topography. If the absolute topography is
known, one can derive the total surface geostrophic
current from the spatial gradients.
Only a relative dynamic topography can be computed
from in situ hydrographic sections [Wyrtki, 1975; Levitus,
1982]. Generally, it assumed that the currents at some
reference level are zero so that the relative topography is
identical to the absolute topography. However, WG80
point out that this assumption is often not correct. One
specific example is the Antarctic Circumpolar Current
(ACC), which extends almost to the ocean floor. Because
of this, WG80 suggested that if the absolute topography
could be recovered precisely from altimetry and a geoid,
then one could combine the absolute and relative
topography measurements to map sub-surface geostrophic
currents with greater detail than has ever been available.
Mapping the absolute dynamic topography with radar
altimetry was a primary goal of the TOPEX/POSEIDON
(T/P) mission, which was launched in August 1992. It
was recognized well before the satellite was launched that
significant improvements in global gravity field models
were necessary to meet this goal both in terms of the
geoid as a reference surface and as an input to the orbit
determination for the satellite itself. A major effort was
undertaken to improve existing gravity models by
reprocessing existing satellite tracking data as well as
new tracking data [Nerem et al., 1994; Tapley et al.,
1996]. The latest such model, the Goddard Earth Gravity
Model 1996 (EGM96) [Lemoine et al., 1998], included
information from all available surface gravity and ocean
altimeter data as well as the satellite tracking data.
Although these advances were significant, the estimated
errors in the marine geoid models were still at the 20 cm
level or worse, at length scales of several hundred km
[Stammer and Wunsch, 1994; Tapley et al., 1994]. As a
consequence, even the largest scale ocean circulation
features could not be accurately resolved, even though
the T/P altimeter is able to measure SSH to an accuracy
approaching 2 cm at the same scale for a period of over a
2. The GRACE gravity model
The Gravity Recovery and Climate Experiment
Mission (GRACE) was designed to help unravel global
climatic issues by enabling a better understanding of
ocean surface currents and heat transport, measuring
changes in sea-floor pressure, observing mass of changes
in the oceans, and by monitoring changes in the storage
of water and snow on the continents. It is expected that
GRACE will provide a minimum of a ten-fold increase
in the accuracy of the Earth’s gravity model to
degree/order 70.
A solution for the Earth’s mean gravity field was
determined from 111 days of GRACE data spanning the
months of April through November 2002 during the
commissioning phase of the mission [Tapley et al.,
2003]. An important consideration in the generation of
this field is that no other information was included as
part of the solution; no a priori constraint, no other
satellite information and no surface gravity information
is used. It is especially important to note that no satellite
altimeter data was used; the preliminary geoid is free
from any sea surface topography signal in the altimeter
measurements. This assumption is not satisfied in any of
the better earlier geoid models. Errors in previous
gravity field models were the result of using data from
multiple sources with varying accuracy and incomplete
geographic coverage. Various portions of the earth have
no gravity measurements and, if the satellite altimeter
data is excluded, neither do major areas of the oceans.
Copyright 2004 American Geophysical Union. Further reproduction or electronic distribution is not permitted.
This deficiency is removed with the GRACE data.
The data set collected by GRACE is global in coverage,
homogeneous in distribution and of very high accuracy.
For spatial, scales as small as 200 km, the GRACE data
used to develop the GGM01 model has improved our
knowledge of the gravity model by an order of magnitude
over the knowledge obtained using over 30 years of
tracking to geodetic satellites such as those used in, for
example, EGM96. When the geoid developed with the
GGM01 model is compared with the geoid from EGM96,
there are major differences over the polar regions and
over mountainous land areas. Over the oceans, there are
20 cm level differences in the regions of the western
boundary currents, the equatorial currents and the artic
circumpolar currents.
3. Results and Discussion
The gravity field coefficients from the GGM01
model to degree and order 90 were used to compute a 1°
gridded map of geoid height relative to the T/P reference
ellipsoid. In order to compute the dynamic topography, a
comparable gridded map of SSH from altimetry is also
needed. We used a gridded mean sea surface (MSS)
model computed from a combination of altimeter
satellites, including the Geosat Geodetic and Exact Repeat
Missions, ERS-1 and -2 (including 186-day repeat orbits),
and TOPEX/POSEIDON [Tapley and Kim, 2000].
However, the spatial resolution of the MSS model is
equivalent to degree/order 8640 (1/24°), compared to
degree/order 90 for the geoid. The MSS therefore
contains the geoid signal to degree/order 90, the geoid
signal above degree/order 90, and the mean dynamic
topography signal. Since the geoid signal above
degree/order 90 can be of the order of several meters,
especially around seamounts and trenches, one needs to
smooth the MSS to a level comparable to the geoid before
differencing the two, or the short-wavelength geoid signal
will obscure any real topography signal.
In this analysis, we have decomposed the MSS into
spherical harmonics to degree/order 90, so as to be
comparable to the GRACE geoid. We have found this is a
better method of filtering than interpolating the highresolution MSS to the grid points, then smoothing,
because it is less sensitive to the locally large shortwavelength signals in the MSS. However, because the
MSS is only defined over the ocean and between latitudes
82°S and 82°N, the missing grid values must be filled
with a reasonable proxy in order to compute spherical
harmonic coefficients. We have filled in the missing grids
over land and poleward of 82° with the geoid computed
from EGM96 to degree/order 360. The spherical
harmonic coefficients of the MSS model were computed
to degree/order 90 and then used to compute the heights at
the same 1° grid locations as the GGM01 geoid. Then, the
dynamic topography was computed by differencing the
gridded values (MSS – geoid) and masking land based on
a land/sea database. An identical procedure was used to
create the results with the EGM96 geoid.
In order to assess the improvement in topography and
ocean circulation due to the GGM01 geoid, we compare
the grids with a relative topography computed from data
in the World Ocean Atlas 2001 (WOA01) [Stephens et
al., 2002] to 3000 m and 4000 m depth by V. Zlotnicki
[personal communication, 2003]. The temperature and
salinity data in this WOA01 are interpolated to a 1° grid
using weighted averages where the weights (WS) are
È Ê ˆ2˘
W S = expÍ-4Á ˜ ˙ ,
ÍÎ Ë S ¯ ˙˚
r is the distance (in km) from the center of the grid for
which the average is desired to the nth grid around it and
RS is the averaging radius. For WOA01, the interpolation
† as a 3-step iteration with RS = 888 km, 666 km,
was done
and 444 km. In order that the topography maps from the
altimetry and geoid models have comparable
smoothness, we have applied the same weighted average
(1), but with R S = 555 km. This will also reduce shortwavelength variations that appear in the altimeter
topography maps due to inconsistency between the MSS
over the ocean and geoid model over land, as well as
residual errors in some high degree/order GGM01
gravity field coefficients.
We compute the zonal and meridional circulation
from the all the topography maps using forwardbackward differences between adjacent grids and
accounting for the change in the area of an equi-angle
grid away from the equator. Values were computed
except within ± 2° of the equator, where values are set to
no data. The circulation for the WOA01 maps was first
computed relative to 4000 m depth. If no value existed in
a grid (because the ocean depth was less than 4000 m),
the current from the 3000 m map was used, if available.
Circulation maps are very useful to evaluate
improvement, since small changes in the geoid can lead
to significant changes in the circulation, especially in the
tropics. Figure 1 shows the zonal velocities (based on the
north-south gradient), while Figure 2 shows the
meridional currents (based on the east-west gradient).
The GRACE topography shows all major zonal
geostrophic currents (Figure 1), especially those in the
tropics. Qualitatively, the locations and magnitudes are
very similar to those from the WOA01 maps. The
EGM96 topography, however, shows no strong currents
in the tropics, and a significantly weaker Kuroshio
Extension and Gulf Stream. The ACC in the EGM96
map is also noticeably weaker than that in the GRACE
map. The GRACE and Levitus meridional currents,
albeit weaker, are also in good agreement. Both show
similar drifts toward the equator in the central Pacific
and Atlantic, while there are equatorward drifts in the
Indian Ocean and western Pacific. The northward
flowing Kuroshio and Gulf Stream are evident in both
the GRACE and Levitus maps, although the GRACE
map shows better detail closer to land due to the depth
one has to use for the WOA01 maps. Although the
EGM96 topography shows similar drifts near the equator
and in the Kuroshio, it does not show the meridional
motion of the Gulf Stream at all.
Copyright 2004 American Geophysical Union. Further reproduction or electronic distribution is not permitted.
Table 1 lists statistics of a comparison of the GRACE
and EGM96 velocity maps with the WOA01 maps. The
RMS of the difference between the WOA01 and GRACE
maps is significantly lower than when the EGM96 geoid
is used for both zonal and meridional currents. Note that
the RMS of the total zonal currents using the GRACE
geoid is 7.2 cm/sec, which is nearly as high as the RMS
difference between EGM96 and WOA01. The RMS
difference between the EGM96 and WOA01 maps is
higher than the RMS of total meridional current (4.4
cm/sec). The correlation between the WOA01 and
GRACE maps is also significantly higher than between
the EGM96 and WOA01 maps.
4. Conclusions
At wavelengths of 500 km and longer, the GRACE
GGM01 Model produces a significantly better marine
geoid than any previous model. This conclusion follows
from evaluating the geostrophic currents determined by
combining the model with a mean sea surface from
altimetry. The agreement with currents computed from a
traditional hydrographic map is very close, which
suggests that one of the primary missions of the
TOPEX/POSEIDON mission, to determine the absolute
dynamic ocean topography, may soon be met.
Because of the filtering necessary when using sparse
data, global hydrographic maps will typically not have as
fine a spatial resolution as one computed from altimetry
and an eventual geoid based on GRACE tracking and
surface gravity measurements. Thus, it may soon be
possible to study geostrophic circulation on an even finer
scale than has been possible with in situ measurements.
By combining the absolute topography from altimetry and
GRACE with in situ hydrographic data, oceanographers
may also be able to map sub-surface currents on a global
scale for the first time [e.g., WG80].
The geoid model presented here is based on data
available during the commissioning phase of GRACE,
with varying degrees of accuracy. Several problems have
been detected during the commissioning phase and have
been corrected in the new data processing. A complete reprocessing of the previous data is underway. It is expected
that this re-processing will lead to significant
improvements over this initial model. The expected
accuracy of final GRACE geoids is expected to be a few
mm out to degree/order 70. However, the MSS is
probably only accurate to a few cm at the same scale.
Thus, the limiting factor in precisely determining absolute
topography from space may once again be the altimetry
measurement. Further research will be required to
quantify this error and to reduce it in order to benefit
completely from the anticipated accuracy of future
GRACE geoids.
Acknowledgements. We would like to thank V. Zlotnicki for many
useful discussions related to calculating dynamic topography and
geostrophic circulation, and for calculating the relative topography
maps. This research was supported by the NASA ESSP Pathfinder
program under NASA Contract # NAS5-97213.
Lemoine, F. G., et al., The development of the joint NASA GSFC and
NIMA geopotential model EGM96, NASA TM-1998-206861,
Levitus, S., Climatological atlas of the world ocean, NOAA Prof. Pap.
13, U. S. Govt. Print. Off. Washington, D.C., 1982.
Nerem, R. S., F. J. Lerch, J. A. Marshall, E. C. Pavlis, B. H. Putney, B.
D. Tapley, R. J. Eanes, J. C. Ries, B. E. Schutz, C. K. Shum, M.
M. Watkins, J. C. Chan, S. M. Klosko, S. B. Luthcke, G. B. Patel,
N. K. Pavlis, R. G. Williamson, R. H. Rapp, R. Biancale, and F.
Nöuel, Gravity model development for TOPEX/POSEIDON:
Joint Gravity Model-1 and 2, J. Geophys. Res., 99, 24421–24447,
Picaut, J., A. J. Busalacchi, M. J. McPhaden, and B. Camusat,
Validation of the geostrophic method for estimating zonal
currents at the equator from Geosat altimeter data, J. Geophys.
Res., 95, 3015-3024, 1990
Stammer, D. and C. Wunsch, Preliminary assessment of the accuracy
and precision of theTOPEX/POSEIDON altimeter with respect to
the large-scale ocean circulation, J. Geophys. Res., 99, 2458424604, 1994.
Stephens, J. i., T. P. Antonov, T. P. Boyer, M. E. Conkright, R. A.
Locarnini, T. D. O’Brien, and H. E. Garcia, World Ocean Atlas
2001, Vol. 1: Temperatures, S. Levitus, ed., NOAA Atlas,
NESDIS 49, U. S. Government Printing Office, Wash., D.C., 176
pp., 2002.
Tapley, B. D., D. P. Chambers, C. K. Shum, R. J. Eanes, J. C. Ries,
and R. H. Stewart, Accuracy assessment of the large-scale
dynamic ocean topography from TOPEX/POSEIDON altimetry,
J. Geophys. Res., 99, 24605-24617, 1994.
Tapley, B. D., M. M. Watkins, J. C. Ries, G. W. Davis, R. J. Eanes, S.
R. Poole, H. J. Rim, B. E. Schutz, C. K. Shum, R. S. Nerem, F. J.
Lerch, J. A. Marshall, S. M. Klosko, N. K. Pavlis, and R. G.
Williamson, The joint gravity model 3, J. Geophys. Res., 101,
28029-28049, 1996.
Tapley, B. D. and M. C. Kim, Applications to Geodesy, Satellite
Altimetry and Earth Science, International Geophysics Series,
Vol. 69, L. L. Fu and A. Cazanave (eds.), Academic Press, 371406, 2000.
Tapley, B. D., S. Bettadpur, M. Watkins and Ch. Reigber, Early results
from the Gravity Recover and Climate Experiment (GRACE), in
preparation, 2003.
Wunsch, C., and E. M. Gaposchkin, On using satellite altimetry to
determine the general circulation of the ocean with application to
geoid improvement, Rev. Geophys., 18, 725-745, 1980.
National Research Council, Satellite Gravity and the Geosphere:
Contributions to the Study of the Solid Earth and its Fluid
Envelope, National Academy Press, Washington, D.C., 1997.
Table 1. Global statistics of difference in velocity maps computed with geoid
models and that computed with WOA01 to 4000 m depth. (mean and RMS in cm/s)
zonal component
meridional component
Copyright 2004 American Geophysical Union. Further reproduction or electronic distribution is not permitted.
Figure 1. Zonal geostrophic currents determined
from the EGM96 geoid (top), a preliminary
GRACE geoid (middle), and from the WOA01
hydrographic data (3000-4000 m). Positive
currents are toward the east.
Figure 2. Meridional geostrophic currents
determined from the EGM96 geoid (top), a
preliminary GRACE geoid (middle), and from
the WOA01 hydrographic data (3000-4000 m).
Positive currents are toward the north

Similar documents


Report this document