PREVIOUS VERSION (8/2006) REMOVED
PLEASE RE-DOWNLOAD *ALL* DATA

JUMP TO WHAT IS NEW

WHAT IS 'EQUIVALENT WATER THICKNESS'?

The observed monthly changes in gravity are caused by monthly changes in mass. The mass changes can be thought of as concentrated in a very thin layer of water at the surface, whose thickness changes. In reality, much of the monthly change in gravity is indeed caused by changes in water storage in hydrologic reservoirs, by moving ocean, atmospheric and cryospheric masses, and by exchanges among these reservoirs, whose vertical extent is much smaller than the radius of the Earth.

The mass of the atmosphere is removed during processing using ECMWF fields, so these grids do not reflect atmospheric variability, except for errors in ECMWF.

Similarly, an ocean model is used to remove high frequency, wind and pressure-driven ocean motions. The values of that model are removed during processing. The resulting gravity fields would not reflect ocean variability if the model were perfect. To use these results over the oceans, the GRACE solutions provided here have the monthly averaged ocean model grids added back. This is the reason we provide OCEAN and LAND grids separately. The GRACE data are the same but the ocean grids have the ocean + atmosphere model added back while the land grids assume the atmosphere model removed was perfect.

Here we present the changes in equivalent water thickness. The method is explained in Wahr et al., 1998. However, the specific processing used here is due to Chambers 2006.

DATA VERSIONS

GRACE is a first-of-a-kind mission, so not surprisingly, revisions to the data processing are more frequent than for more mature satellite measurements.

Three centers are part of the GRACE Ground System and generate level 2 data (spherical harmonic fields): CSR (U. Texas / Center for Space Research); GFZ (GeoForschungsZentrum Potsdam); and JPL (Jet Propulsion Laboratory).

In addition, significant effort has gone into removing 'striping', an error source in the data probably due to unmodeled fast mass changes with periods lower than 1 month (tides, ocean, hydrology) that looks like near N-S  stripes in monthly maps.

At present (5/2007) RL04 data are the most current. Grids from these datasets are available

Some months with poor data distribution in the CSR solutions seem to have been constrained (unconfirmed).

Please download ALL MONTHS from these new solutions and discard previous versions.

Spherical harmonic coefficients of (degree,order) (2,0)  sometimes disagree with those from satellite laser ranging.  The RL02 and later solutions from all centers have more reliable (2,0) coefficients from GRACE than RL01. However, Sean Swenson (U. Colorado) pointed out a large semiannual signal over Antarctica in the (2,0) coefficients of all processing centers.  Don Chambers and colleagues at the U. of Texas then pinpointed it as a 161 day fluctuation, the GRACE alias for the S2 tide. Because of this known error in the (2,0) coefficient, all (2,0) coefficients used in the grids provided here have been replaced by those from SLR (satellite laser ranging).

GRACE cannot retrieve spherical harmonic coefficients of degree 1, proportional to the position of the Earth’s geocenter relative to an Earth-fixed reference frame. Here an annual harmonic  fit to the degree 1 coefficients from Chen (1999) is used instead (see Chambers, 2006a, for details)

All RL04 solutions are major improvement over previous version, as they correct several problems found at all levels of processing, include new background fields, etc.

POST-GLACIAL REBOUND (GLACIAL ISOSTATIC ADJUSTMENT)

The data provided here have NOT been corrected by any PGR model.  Please, see our separate PGR discussion page.

DESTRIPING and SMOOTHING

A source of error, whose telltale signature are N-S stripes, probably caused by aliasing of imperfectly modelled tides, wind-driven oceans, hydrology etc with periods shorter than 1 month plus instrumental effects. While the first two are modeled out in the GRACE processing, the models are imperfect which leaves aliasing energy and causes the striping error.  Swenson and Wahr (2006) observed a peculiar property of the spherical harmonic coefficients associated with the striping, and designed a class of filters to remove the problem.

The GRACE satellites fly at over 400 km altitude. The gravity field weakens with altitude, and short wavelengths attenuate more than longer ones. As a consequence it is necessary to smooth short wavelengths to recover the set of masses on the Earth surface that cause the
gravity field seen by GRACE at its altitude.  To reduce this source of noise, a spatial averaging smoother is applied.

The grids provided here are an implementation of the carefully calibrated combination of destriping and smoothing of Chambers (2006b), who calibrated his results against sea surface height corrected for climatological steric expansion and contraction. For this set, the lower 11x11 set of harmonics was left unchanged, all higher coefficients were adjusted.

Users may need to be aware that the monthly grids have higher errors when the orbit is near exact repeat. Such months include July_December 2004, although the CSR grids for these months are more usuable.

Three smoothing values are given, expressed as the half-width of the equivalent gaussian smoother: 400km, 500km, 750 km.  The 400km grids are fairly noisy relative to the weak ocean signals, but not so relative to the larger land hydrologic signals. They are provided for users who would average the pixels over larger areas, for example, a hydrologic basin.

LAND CONTAMINATION OF OCEAN SIGNALS

Ocean signals are typically weaker than land signals, by factors of 2 or 3. This is true both on seasonal and interannual time scales.  High latitude ocean signals are stronger than low latitude ocean signals. The spatial filters (400-750 km gaussian averages) used to decrease high wavenumber error also imply that a a value at an 'ocean pixel' within, say, 200 km of land, will include part of that land signal. If that land signal is very large it may overwhelm the ocean value.  To help avoid this error, we provide land contamination masks that we recommend be applied when using the ocean data.  An example of such land mask is given in the following figure of water thickness trends (cm/yr) derived from these data:

 With the data below, we also provide several such landmasks, which differ by the relative strength of the gaussian filter that one wishes to discard.

<>BROWSE IMAGES and NUMERIC DATA

Browse images are provided for all the datasets.

The time-average over the time period 1/2003 to 12/2005 of that set has been removed from the data.

The gridded data and browse images are publicly available at

ftp://podaac.jpl.nasa.gov/pub/tellus/monthly_mass_grids/chambers-destripe-RL04-200711/

UNITS and FORMAT

The data are now provided in three formats:

 GEOTIFF, suitable for GIS software
 NETCDF, suitable for automatic ingestion into several software packages.
 ASCII, a plain text format described below.

The simple ASCII format are lines with
longitude  latitude  value
longitude  latitude  value
...

There are 17 header lines preceding the first lat-lon-value data record. The contents are self-describing. For example, file

HDR ATFORMAT015 NHEADER_ROWS=17
HDR FILENAME=GRC_JPL_RL04_DPC_LND_500_200610.txt
HDR VARIABLE=WATER THICKNESS FROM GRACE
HDR UNIT=CMeqH2O
HDR TIME MEAN REMOVED=2003-2006
HDR TIME=200610
HDR LON1=0.5. LAT1=-89.5. DLON=1. DLAT=1. NLONS=360 NLATS=180
HDR CENTER/RELEASE=JPL_RL04
HDR POSTPROCESS=DESTRIPE FILTER, DPC_20070411
HDR POSTPROCESS=NONE
HDR FILTER=GAUSSIAN
HDR FILTERPARAMS WIDTH=500km
HDR FILTERPARAMS MAXDEG=60
HDR MASK=LAND ONLY PIXELS (VZ-2006-11-22-ETOPO5)
HDR INPUT DATA FILE=JPL_RL04_land_500km_200610.txt
HDR FORMAT=f6.1,1x,f5.1,1x,f8.3
HDR DATE=10may2007
   0.5 -89.5    5.078
   1.5 -89.5    5.069
   2.5 -89.5    5.071
   3.5 -89.5    5.063
   4.5 -89.5    5.064
   5.5 -89.5    5.057
   6.5 -89.5    5.058
   7.5 -89.5    5.058

If reading in FORTRAN 77, for example,

REAL GRC(360,180), RLON1, RLAT1, GRC1
INTEGER ILON1, ILAT1
OPEN (LUN, ...
DO k=1, 17
  READ (LUN, *, END=499 ..., ERR=...)
END DO DO k=1, 360*180
  READ (LUN, *, END=499 ..., ERR=...) RLON1, RLAT1, GRC1
  ilon1  = (RLON1 - 0.5) + 1
  ilat1  = (RLAT1 - 89.5) + 1
  GRC(ILON1, ILAT1)  = GRC1
END DO
499 CONTINUE
  ...
END

TIME SPANS OF EACH MONTHLY SOLUTION

'Monthly' is used somewhat loosely: please see the table of actual data days used for each 'monthly' solution. It is important to note that the GFZ solutions (not given here) use slightly different days for the same approximate months. The difference is due to data editing.

TIME AVERAGE REMOVED FROM MONTHLY SOLUTIONS

Each monthly grid here represents the difference the masses for that month, and the average over 2003-2005. If you compare against other data or model, it is critical that anomalies from the same time-average be compared; this is simply done by removing the average over the chosen period from any set of grids, including those provided here.

CITATION

When using these data, please include this phrase in the acknowledgements

GRACE data were processed by D. P. Chambers, supported by the NASA Earth Science REASoN GRACE Project, and are available at http://grace.jpl.nasa.gov

and cite:

Chambers, D.P.: Evaluation of New GRACE Time-Variable Gravity Data over the Ocean. Geophys. Res. Lett., 33(17), LI7603, 2006.

REFERENCES used above:

Chambers, D. P: Observing seasonal steric sea level variations with GRACE and satellite altimetry, J. Geophys. Res., 111 (C3), C03010, 10.1029/2005JC002914, 2006.

Cheng, M. and Tapley, B.D.: Variations in the Earth's oblateness during the past 28 years, J. Geophys Res v109, B9, 2004

Swenson, S. C. and J. Wahr, Post-processing removal of correlated errors in GRACE data, Geophys. Res. Lett., 33, L08402, doi:10.1029/2005GL025285, 2006.

Wahr, J., M. Molenaar, and F. Bryan, Time-variability of the Earth’s gravity field: Hydrological and oceanic effects and their possible detection using GRACE, J. Geophys. Res., 103, 32,205–30,229, 1998.