GRACE AOD1B RL04 Quality Assurance

At GFZ Potsdam the GRACE Atmosphere and Ocean De-aliasing Level-1B (AOD1B) products are generated on a routine basis for the period January 1, 2001 until today (current status January 2, 2012) to be used for correction of short-term non tidal atmospheric and oceanic mass variations for GRACE (and CHAMP). The latest release 04 (RL04) is based on 6-hourly ECMWF analysis data and output from the baroclinic ocean model for circulation and tides (OMCT). Details on OMCT outputs can be found at here. Additionally, for a consistent processing of SLR data the RL04 time series has been prolongated back to January 1, 1976 based on ERA40 data and made available to the users on January 19, 2009.

All products for 1976 till today are stored as daily gnu-zipped ascii files (AOD1B_YYYY_MM_DD_X_04.asc.gz) at the GRACE archives at ISDC and PO.DAAC and contain spherical harmonic coefficients up to degree and order 100 for four 6-hourly time stamps (0, 6, 12 and 18 hours) and for each time stamp four different data types describing gravity variabilities w.r.t. a 2001+2002 mean field are provided:

ATM: variability of the vertical integrated atmosphere
OCN: variability of the OMCT ocean model output
GLO: variability of the global combination of atmosphere and ocean
OBA: variability of the OMCT ocean bottom pressure

A typical example for the variability in terms of geoid height for November 1, 2007 00:00 is shown in the following figure.

Figure 1: Geoid height variability [mm] for AOD1B RL04 components "atm", "ocn", "glo" and "oba" on November 1, 2007 00:00.

Details on the GRACE RL04 products as well as on the precursor releases can be found in the GRACE AOD1B Product Description Document for Product Releases 01 to 04 (Rev. 3.1, April 13, 2007) which can be downloaded here (1.7 MB). Details on the additional time series for 1976-2000 based on ERA40 are described in a Scientific Technical Report (4.5 MB).

In order to check the quality of the GRACE mission period products different plots are automatically generated and shown on this page (plots for the period 1976-2008 are included in the STR0812). First, the minimum, maximum, mean and wRMS (cosine latitude weighted RMS) geoid heigth variability for all four data types and for the complete production period (2001-today) is derived from spherical harmonic coefficients of degree and order 0 to 100. Second, the individual degree 0 to 2 spherical harmonics 6-hourly time series for the "atm", "ocn", "glo" and "oba" variabilities are shown in colour and high pass filtered (after subtraction of the daily mean) in black.

top | C00 | C10 | C11 | S11 | C20 | C21 | C22 | S21 | S22 | geoid

Generally, for a "nominal" AOD1B product, the

mean geoid "atm" variability should be close to zero (due to subtraction of a long-term 2001+2002 mean) and should show an annual signal.
mean geoid "ocn" variability must be zero because (in contrast to AOD1B RL03, see AOD1B PDD chapter 5) a mass conserving approach has been used in the RL04 OMCT runs. This is also reflected in the very small amplitude of the "ocn" C00 time series below.
minimum and maximum "atm" and "ocn" variabilities should be generally below 20-25 mm geoid height.
minimum and maximum "glo" geoid height variability should be slightly smaller than for "atm" due to the quasi compensation of "atm" plus "ocn" over the oceans.
"oba" results should show smaller minimum and maximum geoid height variabilities compared to "glo" because for "oba" the land pixels have been set to zero (and therefore have no variablity).
"oba" mean should show a higher amplitude than the "glo" time series due to the use of surface pressure instead of vertical integrated pressure.
weigthed rms for all four components should be about 2-4 mm geoid height.
high pass filtered Cnm and Snm should show no jumps in amplitude (as visible in wrong AOD1B RL04 C22 and S22 time series for June 23, 2006 until September 20, 2007).
Important Note
If you downloaded AOD1B RL04 products for the period June 23, 2006 until September 20, 2007 prior to November 30, 2007 please be sure to download these files again because the oceanic contributions to these products are wrong (by mistake twice reduction of atmospheric S2 tide from surface pressure data used to force the OMCT model). The correct AOD1B RL04 products for this range of dates have to have a product creation date later than November 2007 within the header information, e.g.
PRODUCT CREATE START TIME(UTC): 2007-11-08 10:51:24.000
PRODUCT CREATE END TIME(UTC) : 2007-11-08 13:19:25.000
The influence of the wrong AOD1B products on the GAA/B/C/D products is negligible, thus these products need only to be redistributed if the number of days to generate the GSM-products have been changed. The influence on the gravity field result is unfortunately drastic (especially over the oceans). Therefore the SDS centers have reprocessed the corresponding GSM products. Please read the SDS newsletters and the L2 Release Notes (both available at the GRACE archives) regularly.

Interestingly, the atmospheric mass is increasing till 2005 by about 1.9 10^14 kg/year and is then decreasing by about 4.7 10^14 kg/year (see "atm" C00 time series). But, as the AOD1B degree 0 and 1 terms are - as for all other background models - not used (fixed to 1 and 0) for GRACE gravity field determination, this does not influence the quality of the GRACE level-2 products.

top | C00 | C10 | C11 | S11 | C20 | C21 | C22 | S21 | S22 | geoid