In processing the data for the tide height corrections, we found a surprising number of incidents when the fix time went backwards - 404 instances in the 567 log files that have been submitted to date.
On examining this, the vast proportion of the instances are just a step back of a second or two. This could be the case when the receiver uses a different set of satellites to calculate the fix - particularly if there is a big change in the satllites used, for example if part of the sky is obscured. This is not an issue for TeamSurv, as the error is not lartge enough to be significant, and most navigation systems will also be untroubled by this.
Of the others, the Lowrance XLS27C would occasionally jump backwards in time by a significant amount for a minute or so, before jumping back to the correct time. However it was well behaved enough to flag these fixes as invalid, even though the position appeared OK. As the fixes are flagged as invalid, any application or device receiving them should just ignore them, which is what we do.
Another issue is with the rollover from one day to the next. A Raymarine Raystar GPS doesn't update the date for the first RMC sentence of the new day, but this is corrected on the next message, so this transient issue is unlikely to be too significant.
We also have a couple of loggers where the GPS jumps back in time, but the NMEA output is still flagged as valid. With one GPS, the time resets to 00:00:00 1/1/2003 and starts counting up from there, but with the other the date is unchanged, and the time just resets to 00:00:00. This is totally unacceptable, and users should avoid these products. Because we are not doing things in real time, we can detect and discard this data. But most marine navigation systems do not have a memory of past records, so it will just accept these as valid. This will play havoc with ARPA and AIS targets, as well as with any route calculations, tide height or tidal stream calculations. The effects could be anything from a major nuisance to outright dangerous. The two loggers with these items haven't given us the make and model of their GPS units, but we will ask them and make this available when we have it.