Anomalies in the DMSP F-16 input data for the TMPA-RT created streaks of precipitation over open-ocean regions. We have now deleted faulty data and rerun the TMPA-RT for the period: 18 December 2016 at 09 UTC to 19 December 2016 at 18 UTC. Please re-pull the affected data files. There appears to be no corresponding issue with the IMERG data files due to different quality control routines.
UPDATE 9/27/16: January 2016 IMERG has been regenerated and posted. This was necessary due to an error that omitted the gauge analysis in the Final products. Also, 3B42/3B43 have been recomputed and posted. Let me know if you have any further questions. --- PPS has removed the January 2016 GPM IMERG and January through June 2016 3B42/3B43 TMPA products from our ftp archive: ftp://arthurhou.pps.eosdis.nasa.gov/ and STORM: https://storm.pps.eosdis.nasa.gov/storm/ The GPM science team discovered that the IMERG final products for Jan 2016 and the Jan-Jun 2016 TMPA 3B42/3B43 products for which we
FNMOC is behind in providing the SSMIS data used in IMERG. Therefore, the Early and possibly Late IMERG products will contain less satellite estimates and be of somewhat lesser quality. FNMOC stopped providing data around 00 UTC today (8 September 2016), meaning there has been no data available from the three operational SSMIS sensors to the TMPA-RT datasets. As a result, more lower-quality IR-based estimates are being used. In addition, if there are dropouts in the IR data, there is a higher probability that gaps will appear in the combined product, 3B42RT. We will update the situation when
The Washington, DC area is about to get hit with a record-level snow storm, with attendant disruptions to work schedules, travel, communications, and power. If you find that the TMPA-RT products drop out, it is safe to assume that some combination of these effects is preventing production, and we will work to restore service when that is possible.
With the turnover of the year to 2016, the suite of 3B40RT, 3B41RT, and 3B42RT failed to process. Despite the holiday weekend, this was traced to a very old quality control on the date/time, fixed, and retrospectively processed to give a complete time series. Thanks to Erich Stocker and David Bolvin for this effort. The affected date/times are 160101 00UTC through 160102 12UTC.