Jason-2 reprocessing

Since the launch of the Jason-2 satellite on June 20, 2008, the GDR (Geophysical Data Record) data were distributed in version T. For the first time, this year, a whole mission reprocessing occurred for Jason-2 altimetric data set. All cycles from 1 to 145 (covering the period from July 12, 2008 to June 18, 2012) were reprocessed into a homogeneous standard (GDR-D version) and added to the current production time series. Note that the operational version of Jason-2 GDR switched to standard D from cycle 146 onwards (for IGDR, from cycle 150 onwards; see operational news: switch to GDR-D version).

The global report presented here deals with the complete reprocessed period (cycles 1 to 145) of the Jason-2 mission, thanks to comparison with previous Jason-2 GDR-T standard, as well as comparison with Jason-1 and Envisat data. It also contains the impact of the reprocessing on the mean sea level trend. (see complete report of Jason-2 reprocessing file).

Preliminary comparison:

A first overview of the impact of the GDR-D version of Jason-2 altimeter system over ocean was performed over the formation flight phase (cycles 001 to 020 from July 12, 2008 to January 26, 2009). [ref: S. Philipps, H. Roinard and M. Ablain, 2012: "Jason-2 reprocessing impact on ocean data (cycles 001 to 020). Comparison of Jason-2 Gdr-D with Gdr-T, as well as with Jason-1 Gdr-C." JA2_GDR_D_validation_report_cycles1to20_V1_1.pdf]

In terms of available and valid data:

Data availability of GDR-D products is equivalent to the data availability of GDR-T products. For some cycles, data which were missing in GDR-T version are present in GDR-D version. Percentage of invalidated data is similar between GDR-T and GDR-D. Nevertheless, GDR-D products have more edited data likely due to the use of MQE criterium during the 20Hz to 1Hz compression.

In terms of performance at cross-overs:

GDR-D data improve the coherence of ascending / descending SSH differences as geographically correlated patterns are reduced.

Mean in cm of ascending/descending differences of SSH at crossovers, using GDR-T (left) or GDR-D (right) standards. Credits CNES/CLS.

Difference of SSH variance at crossovers between GDR-D and GDR-T (open ocean selection: 50°S < latitude < 50°N, bathymetry <-1000m, and oceanic variability <20cm). Credits CNES/CLS.

The standard deviation of SSH differences is systematically lower for GDR-D than for GDR-T data. The global SSH variance reduction is about 1.7cm², with occasionnally reductions up to 3cm² for a cycle. This variance reduction and a mean closer to zero with reduced correlated patterns are signs of an improved coherence of ascending/descending passes, which is mainly due to the new orbit solution, the use of GOT4.8 tide instead of GOT00 tide,and the modified datation used for GDR-D data.

In terms of Global Mean Sea Level trend:

Difference of MSL trends of Jason-2 computed by separating ascending and descending passes, removing annual and semi-annual signals. Credits CNES/CLS.

For GDR-D, there is a very good improvement as the odd/even pass MSL trend differences is reduced to less than 0.1mm/yr (instead of 0.74mm/yr using GDR-T data set). Note that a near-120-days signal still remains.

Global MSL trends for Jason-2 (2months filtered, radiometer wet troposphere, semi-annual and annual signals removed). Credits CNES/CLS.

The global MSL trend using Jason-2 GDR-D increases by 0.2mm/yr compared to GDR-T Global MSL when using radiometer wet troposphere correction. The main contributor to the trend modification between GDR-D and GDR-T data is the radiometer wet troposphere correction (+0.35mm/yr when using the GDR-D radiometer wet troposphere correction) and to a lesser extent the POE-D orbit standard (-0.1mm/yr when using the POE-D orbit standard).

In terms of Regional Mean Sea Level:

Hemispheric east / west differences up to +/-3mm/yr are observed, mainly related to the change of orbit solution.

Regional MSL difference of trends between GDR-D and GDR-T data. Credits CNES/CLS.

Details about main evolutions:

The main improvement in GDR-D version is its new orbit. This solution very much improves the coherence of ascending/descending SSH differences. It also improves the coherence between Jason-2 and Envisat as there are no longer east/west biases
The pseudo time-tag bias mean (-0.29ms for Jason-2 GDR-T) has been corrected and is now close to zero. Consequently, the small north/south hemispheric differences on Jason-2 SSH crossovers differences are reduced.
Radiometer parameters have been enhanced in coastal regions, and an anomaly in 34GHz channel has been corrected. New AMR calibration coefficients are also used.
The expected modifications with the update of the altimeter characterization file, corrected pole tide solution, altimeter wind speed, global tide model are observed.
GDR-D sea state bias is calculated with a different approach for low sea states. Main differences are observed in these areas (as the editing method has changed). Note that an updated version of SSB model (based on GDR-D data) was presented at OSTST2012.