TOPEX/POSEIDON Altimeter Gridded Sea Level Anomalies and Deviations

Bob Cheney
Russ Agreen
John Lillibridge
Nancy Doyle
Laury Miller

National Ocean Service, NOAA (N/OES11)
Silver Spring, MD 20910

The data are here.

This new "read_me" file documents some changes to our T/P analysis.

As of April 1996:

the original JGM-2 orbit was replaced with JGM-3, reducing radial orbit error from about 3.5 to 2.5 cm rms.
data gaps of up to 10 seconds were filled by interpolation, resulting in a more complete global data set.
the local inverted barometer correction was augmented by a non-local correction (usually less than 1 cm).
sea level time series were expressed as deviations relative to the 3-year mean, Jan 1993-96, rather than differences relative to the first observation.

As of July 1996:

Topex oscillator error (12-15 cm) was corrected.
Topex internal calibration (less than 1 cm) was applied.
CSR 3.0 ocean tide (Univ. Texas) was incorporated (instead of Cartwright/Ray with Wagner upgrade).
each Poseidon cycle was adjusted for an offset relative to adjacent Topex cycles (usually less than 1 cm).

The net effect of most of these changes is to reduce the noise in the altimeter sea level time series.

In contrast, the Topex oscillator and internal corrections alter the large-scale, low-frequency sea level signal, having the combined effect of reducing sea level at any given location by about 8 mm per year (relative to the earlier, uncorrected Topex data).

It is recommended that previous versions of the T/P data be discarded.

Bob Cheney

1. SUMMARY

The compressed file, global_2_136.lo_res.Z, is a collinear difference analysis of Topex/Poseidon altimeter data for cycles 2-136 (Oct. 3, 1992 - June 2, 1996). Results are expressed as time series of sea level deviations, with respect to a long-term mean, averaged along 1-degree latitude intervals of the satellite track. Upon request, similar analyses can be made available on a regional basis for 0.2-degree latitude averages.

(a) Headers

Each header gives N latitude and E longitude in degrees for a particular 1-degree latitude segment of the satellite track. The flag gives several pieces of information: - value indicates a descending track (NW to SE) + value indicates an ascending track (SW to NE) 8 in the thousands column indicates that this is a header remaining digits give the equator crossing longitude in degrees For example, the first header above gives a location of 7.00 N, 198.10 E and the 1-degree segment is part of a descending track that crosses the equator at 200.60 E.

(b) Data Records

Following the header is the sea height time series at this location, where the three columns are time (1985 day), H (cm), and n (the number of 1-sec altimeter records averaged to obtain H). For a 1-degree latitude segment, the maximum number of 1-sec records varies from about 21 at the equator to 32 at 60 N. In the example above, the first data record is at time 2833.28, has a height of -7.15 cm, and is an average of 12 1-sec records. The time represents the equator crossing of the pass, rounded to the nearest hundredth of a day. By our definition, 0 = January 1, 1985, 00 hrs, so that 2833.28 corresponds approximately to October 4, 1992, 0700 hrs UTC. The subsequent data records, arranged in chronological order, give the sea level at ~10-day intervals, where each height is the deviation relative to the long-term mean (see next section).

(c) Reference Period

In the present file, each time series is expressed relative to the 3-year mean (Jan 1993-96) at that particular location. As the data set becomes longer, this reference period will be extended, but will always be an integer number of years. In computing the long-term mean, there is no requirement for a minimum number of observations. It is left to the user to check for data gaps which could impact the significance of the mean.

(d) Organization

Latitude/longitude positions of the time series are located along the satellite track, but in the file they are organized into latitude bands scanning from west to east between 0 and 360 E and from south to north between -60 and +60. Ascending passes can be found in the first half of the file, followed by descending passes for the remainder of the global set.

3. PROCESSING METHOD

The analysis includes both Topex and Poseidon data. The geophysical data records (GDRs) are used for most of the mission while the interim GDRs (IGDRs) are used for the most recent cycles to make the data set useful for near real-time applications.

The present file contains:

GDRs for cycles 2-132

IGDRs for cycles 133-136

Orbit precision is about 2.5 cm for the GDRs and 5-10 cm for the IGDRs. JGM-3 orbits are used throughout. The Topex data are those distributed by JPL. The Poseidon data are distributed by CNES, with a small (< 1 cm) offset applied to each cycle based on collinear differences with the adjacent Topex cycles.

The data were processed in the following way using collinear differences:

Standard corrections:
- wet and dry troposphere
- ionosphere
- tides, using the CSR 3.0 model of the University of Texas. Solid earth tide includes pole tide corrections.
- local inverse barometer, based on the instantaneous atmospheric surface pressure
- non-local inverse barometer, based on the global average surface pressure for that particular 10-day cycle
- sea state bias: either Callahan's algorithm for Topex data or Gaspar's 4-parameter correction for Poseidon data.
- OSU91A geoid to reduce errors resulting from slight cross-track gradients within the 2-km collinear band.
Editing Data were excluded if the satellite attitude was greater than 0.3 degrees in cycles 1-8. GDRs have flags that allow exclusion of bad data. Shallow as well as deep water data are allowed. If there are problems with the altimeter sensor corrections, ionospere, sea surface height, or if sea state conditions are invalid, data are rejected. In addition, land or other geophysical conditions cause data to be rejected. Sigma_h, the root mean square of 10/second sea surface heights, is limited to 110 for Topex data and 170 for Poseidon data.
Interpolation Data gaps in the GDRs of up to 10 seconds long are filled by linear interpolation. The ionosphere correction is smoothed over 25 seconds before being applied.
Orbit adjustment No adjustment was made for orbit error.
Altimeter drift The internal calibration correction supplied by NASA Wallops was applied.
Oscillator (clock) drift

The error discovered in July 1996 was corrected using the table provided by NASA Wallops.

TOPEX/POSEIDON	LDEO Climate Group	LDEO

bob@bigbird.grdl.noaa.gov