Level-1 pre-processor does not properly merge l1 segments

[shendric]

Expected behaviour of the Level-1 pre-processor is to merge trajectory data from different input files, if these are connected.

An evalution of CryoSat-2 l1p (cryosat2_pds_ipf1e_v1p1) has revealed that this is not always the case and data gaps arise in areas with a frequent land segments:

The figure shows the (lon, lat) location of all data points in the time_orbit data group for l1p files within a given day. Colors indicate extent of l1p files.

Actions

• Identify and correct issue in l1 pre-processor

[shendric]

Improved with 957228a, but issues still remain. E.g. in the updated plot the trajectory after 31 (was 48 in previous plot) is missing:

Also 13 should be connected to 14

[shendric]

A fix has been found for the missing orbit segment (0ec0073), but there are still some segments that do not connect properly (Example 17, 18, 19 in the figure below)

[shendric]

An additional CryoSat-2 related bug was corrected that led to mixup of TAI and UTC timestamps when checking if l1 polar ocean segments should be merged (74a23c7). The test output is now nominal.

[shendric]

The remaining issue is an inconsistency in the handling of land segments: A land segment of a give length might be included if it occurs within a single source file, but the l1p output might be split into two files if the land segment would occur at the edge of two source files.

In the example above, segments 13 and 14 are split at Nowaja Semlja, while segment 33 (number shown only in previous comment) cut right accros it:

The reason for this behaviour is that the treshold for connecting l1p segments is set to 10 seconds ( max_connected_segment_timedelta_seconds), while the Level-1 pre-processor allows to include 1000 waveforms (= 50 seconds at 20Hz) of land within one segment:

It should be investigated if max_connected_segment_timedelta_seconds should be changend.