Time is recorded by the system  and reported in the data files in two ways.

1)  epochSECS and epochuSECS is the time of the linux machine, however it's set.  We aim to set this to UTC time and sometime maintain it with ntp, there aren't any guarantees this is perfectly accurate though, although generally one would expect it to be within a minute of correct.

2)  The gps data packet includes the GPS time of the week field (ITOW).  This gives milliseconds since the start of the week counting, which happened for GPS at midnight Jan 5, 1980 (the midnight between the 5th and the 6th).  

The post-data processing takes the epochSECS and microseconds, adds them to the matlab time that cooresponds to the unix epoch (Jan 1, 1970), and reports that as matlab serialdate number.  Therefore, if the unix clock is correct, datestr(serialdate) of the processed data in matlab shows the date and time in UTC time.  

If there is uncertaintly about the time, it can be checked by comparing a samples ITOW seconds into the week with the UTC time and date to confirm it's close enough, or calibrate as needed.


In summary, the ITOW is the most accurate time recorded, but requires that you know which week it was recorded in.



Wave-Rider Data
- The wave-rider data downloaded from the CDIP 156 buoy site is in UTC time, the time encoded in the "filename" field of the spectral data files is the UTC time of the first sample of the approximately 1/2 period the spectrum represents.