Lance-MODIS latency and JPSS1 L0 gap handling
|Last Updated||2019-08-28 17:23:00|
NRT JPSS1 PGE595 (which splits sessions based L0 granules into 6 minute granules) is running a new PGEVersion in nrt4 as of the beginning of 2019-08-28. JPSS1 sessions granules cover a bit over an orbit of data and we receive them about every half orbit, so that our 6 minute periods are overlapped by two granules, or three near granule boundaries, and sometimes the earlier granules are missing packets which are present in a later granule. Now if PGE595 produces 6 minute L0 with filesizes that indicate input packets were missing and it did not run with the maximum number of overlapping sessions granules, instead of archiving the six minute L0 with gaps it will fail and auto retry when new overlapping sessions L0 arrives. This should result in some later but more complete VJ1 6 minute products (particularly for sessions granules with data gaps near their trailing end).
And this should help avoid gappy 6 minute L0 inputs causing errors in PGE500 which require manually reprocessing of PGE595 to fix (as happened with the 10:48 granule in nrt3 today), resulting in less latency when this would occur. This change will likely be installed in nrt3 within the next week.
Additional nrt4 latency this morning was caused by import failures due to a software bug installed between 11:30 and 13:30z today.