Repository navigation
Self position
Where the station's own observations put it, against the position it broadcasts. Every reference station tells its users where its antenna is (RTCM 1005/1006); this view takes the station's own GPS observations and broadcast orbits, solves for where they place the antenna, every epoch, and plots the result around that broadcast position.
Open it from Analysis: it is the tab after C/N0 vs elevation.
- Two GPS frequencies. The ionosphere is removed by combining two carriers; a single-frequency station is NOT COMPUTABLE, and waiting will not change that.
- Orbits. From the station's own stream, or — for a station that broadcasts none — from the ephemeris stream. Without them the view says it is waiting for orbits.
- GPS only, deliberately. Each constellation keeps its own time, and mixing them under one receiver clock put a real station 1.8 km from itself.
| Figure | What it says |
|---|---|
| Epochs solved | How many epochs produced a position, of how many in the window |
| Mean offset E/N/U | Where the solutions centre, against the broadcast position |
| Scatter about the mean | How widely they spread — an upper bound, see below |
| Worst code residual | How well the worst epoch's measurements fitted its solution |
| Median apparent motion | The solved velocity. A base standing still reads tens of mm/s |
No verdict. No threshold for a station's own scatter has been established from evidence, so these are figures, not a judgement — and the view says so.
The plot is east and north around the broadcast position, which stays at the centre: a plot that centred itself on the solutions would hide the offset. The latest epoch is a line under it, as a diagnostic.
This is not a survey, and the offset is not the station's error. A position from broadcast orbits and clocks is good to a few metres. Both the offset and the scatter include this solution's own errors:
- The app forms its ionosphere-free combination from the signals a station sends — usually L1 C/A with L5 — against satellite clocks defined for L1/L2 P(Y), and it does not correct the satellite-by-satellite code biases between them. Each satellite then carries an offset of its own, up to a few metres, and the solution moves whenever a satellite rises or sets.
- Measured on one station over three hours, correcting those biases from a published product put the mean within half a metre of the broadcast position and cut the jumps roughly in half. The phone has no way to fetch that product, so the app does not apply it.
So read the scatter as an upper bound on how much the station's own position varies — not as a measurement of it. A small scatter is good news; a large one is not, by itself, a fault in the station.
The apparent motion is unaffected: it comes from the rate of change of the carrier phase, which code biases do not touch. That is the figure that would show an antenna that moved.
REVERSED appears when the station's receiver encodes its range rates with the wrong sign — every satellite the exact negative of what its orbit predicts. The app works out which convention the station uses and the motion is right either way, but a rover that trusts those rates is being misled by them.
Free does the same work during a run — nothing is measured differently between the editions — but has no view that draws it.