Jason Altimetry Data Processing for Enhanced Coastal Sea Level Anomaly Analysis

The paper describes the process of correcting and processing the sea level anomaly (SLA) data obtained from Jason altimetry missions. The corrections include atmospheric effects, geophysical phenomena, and sea-surface state biases. The corrections account for atmospheric effects (wet and dry troposphere, ionosphere, inverse barometer), geophysical phenomena (ocean tides, high frequency atmospheric effects on the ocean) and the sea-surface state (electromagnetic sea-surface bias). At this stage of the processing, we obtain a regional dataset of SLA time series with a temporal resolution of 10 days and a spatial resolution of ~300m along the tracks of each Jason altimetry mission. The computation of a single long-term multi-mission product requires the application of inter-mission biases in order to remove instrument and corrections biases, thus to obtain stable merged sea level time-series. Inter-mission biases are computed during the 'calibration phases' between two consecutive missions (about 3 month-long between Jason-1 and Jason-2 in 2008, and between Jason-2 and Jason-3 in 2016), when both satellites measure the same sea level on the same ground track, with about 1minute time lag. The biases are computed for each of the six studied regions, excluding altimetry points located at less than 10 km from the coast, as well as points where more than 20% of data are missing in the time series. In order to remove the high-frequency variations in the resulting 20-Hz along-track bias values, known to be associated with measurement noise, the data are low-pass filtered (with a 40-km cutoff frequency) and averaged over 1ﾰￗ1ﾰ boxes. The corresponding smoothed 1ﾰￗ1ﾰ bias values are then interpolated at the original 20-Hz along-track altimetry points and applied to the SLAs. The Jason-1/2 inter-mission bias is applied to Jason-2 SLAs first, and then the Jason-2/3 inter-mission bias to Jason-3 SLAs. It is worth mentioning that no orbit error reduction has been applied to the coastal sea level product. This is because such a correction based on differences between ascending and descending satellites tracks due to orbit errors needs to be computed globally. However, the orbit error is supposed to be small at the coast because of its large-scale behavior. Finally, the long-term multi-mission SLA time series are monthly averaged, and a linear trend v and associated error are derived at each 20-Hz along-track point (see section 4.2 for details on the trend calculation). A first version of the processing system described above has been successfully evaluated and validated in. It significantly increases the number of valid SLAs in the coastal zone. As an example, in the Mediterranean Sea,80% of valid sea level data are still available at distances of <2km from coast, compared to distances of several km in the standard X-TRACK product. The resulting regional datasets have a temporal resolution of 10 days and a spatial resolution of ~300m along the tracks of each mission. In order to obtain a stable merged sea level time-series, inter-mission biases are computed during calibration phases between two consecutive missions. The resulting 20-Hz along-track bias values are low-pass filtered and averaged over 1ﾰￗ1ﾰ boxes, and then interpolated at the original 20-Hz along-track altimetry points and applied to the SLAs. The long-term multi-mission SLA time series are monthly averaged, and a linear trend and associated error are derived at each 20-Hz along-track point. The processing system significantly increases the number of valid SLAs in the coastal zone, with 80% of valid sea level data available at distances of <2km from coast in the Mediterranean Sea compared to several km in the standard X-TRACK product. It is worth mentioning that no orbit error reduction has been applied to the coastal sea level product due to the small-scale behavior of orbit error at the coast.