Questions about the daily and hourly file latency graphs

Where do the numbers plotted come from?

The time of file availability is determined by checking the timestamp of each hourly file at the following Data Centers: CDDIS, IGN, SIO, and (for hourly only) BKG. If found at multiple DC's, the earliest timestamp is chosen. The difference between this timestamp and the final epoch of the file gives the displayed latency.

What are the error bars in the 90-day and 3-day average plots?

Standard deviations of the previous 90 days (for daily) or 3 days (for hourly) of data as plotted in the "Latency of each daily/hourly file" plots.

What are "good" numbers?

For daily data, availability within 2 hours is ideal. After 6 hours, the data is not useful to some analysis centers' rapid orbit processes. After 3 days, the data is not useful to most ACs' final orbits or reference frame products. Only the TIGA pilot project allows for usage of data later than this.

The IGS recommends hourly data files be available within 10 minutes past the last epoch in the file. This benefits certain applications such as tropospheric investigations. If this cannot be met due to shortcomings of infrastructure at a given site, it can still benefit other applications which tolerate somewhat higher latencies. Each site should strive to the lowest latency feasible within operating constraints.

But I'm the site operator and I submitted it earlier!

Yes, all users should realize that these times represent the earliest availability at the DCs. This includes any processing or verification time that goes on within the DC.

This page last updated: 13 May 2010
Please send comments about this website to: IGS Central Bureau.