From: Jim Ray (USNO 202-762-1444) Message-Id: <200104121900.PAA21662@maia.usno.navy.mil> Subject: [IGSLEO-6] Re: LEO Workshop summary To: igsleo@igscb.jpl.nasa.gov Date: Thu, 12 Apr 2001 15:00:55 EDT Cc: rweber@luna.tuwien.ac.at (Robert Weber (TU Wien)) Sender: owner-igsleo Precedence: bulk ****************************************************************************** IGS LEO Mail 12 Apr 12:00:59 PDT 2001 Message Number 6 ****************************************************************************** Author: Jim Ray Regarding the nice summary of the LEO workshop, I previously raised a question concerning the stated need for high-rate clocks: > There was collective agreement that IGS > clocks at 30s or 10s would be of great value to these missions. ~~~~~ cut > Summary requirements: > - IGS 10-30sec clocks > - Retrieval of occultation profiles require <0.1mm/sec velocity > determination and ~10cm LEO POD ~~~~~ cut > Currently they provide 30s clocks from 5min > estimates; proposed 1s from 5 min estimates & 1s ground data. Now > that SA is off, a new proposal is to investigate interpolation & > extrapolation for 1s GPS clocks from 5 min estimates & high rate > ground data. ~~~~~ cut > Velocity should be known to 0.1mm/s > (for occultation) and position to ~10cm. I saw these requires again at the recent EGS meeting. My questions concern the specific need for IGS clocks at a rate greater than the present 5-min intervals. Does this include only GPS satellites or also tracking stations? Now that SA is off, why can't this requirement be met adequately by interpolation of the current 5-min clock values? In the position paper by Jim Zumberge and Gerd Gendt for the IGS Network Workshop (July 2000) -- Figure 2 -- I infer that the maximum interpolation error between 5-min points (at 150 sec) would be about 1.7 cm = 57 ps. I seriously doubt that the precision of the IGS combined clocks is really much better than 100 ps at each tabulated epoch, so the interpolation error should not seem to be a great addition. Maybe I'm missing something (e.g., the rate of change of the clocks?) but it just does not seem obvious to me that samples more frequently than 5 minutes really buys you very much. That would seem to be especially true for the newer Block IIR oscillators which are very excellent frequency standards. Thanks and best regards, --Jim