2.3.5. The Results Page

Everything the solver produced, plus the checks that tell you whether to believe it.

2.3.5.1. The Results Grid

One row per baseline.

Column	Meaning
Base / Rover	The two ends. The vector is rover minus base.
State	`ok` if the solver returned a solution, `failed` if it did not, `planned` if it has not been attempted.
Solq	Solution quality: FIX or FLOAT.
Ratio	The ambiguity ratio actually achieved. Compare it against the ratio threshold you set: a fix at 3.1 against a threshold of 3.0 is a very different thing from a fix at 12.
NSat	Satellites used.
Epochs	Epochs used.
RMS	Solution RMS from the solver.
Length	The solved baseline length.
dX / dY / dZ	The baseline vector in ECEF metres.
σ	A single summary standard deviation for the vector, from the solution covariance.
Notes	Per-baseline messages: on a successful solve, warnings such as a generic-ephemeris fallback or a truncated navigation list; on a failure, the reason it failed.

2.3.5.1.1. Row colours

Colour	Meaning
Green	Solved, FIX. This is what you want every row to be.
Orange	Solved, FLOAT. The vector exists but the ambiguities were not resolved; treat it as an order of magnitude weaker than a fix.
Red	Failed. Read the Notes column for the reason.
Plain	Not yet processed.

2.3.5.1.2. Re-solve selected

Select one or more rows and press Re-solve selected to re-run exactly those baselines with the settings currently on the Processing page.

Unlike Run, this deliberately ignores the “not already solved” rule — naming rows means you want those rows redone, and every row you would want to re-solve after changing an option is, by definition, already solved.

The typical loop is: spot an orange row, go to Processing, change the elevation mask or turn off GLONASS ambiguity resolution, come back, select the row, re-solve.

2.3.5.2. Loop Closures

This is the independent check on the whole network, and the reason redundancy matters.

The module builds a fundamental cycle basis over the usable baselines — a spanning tree plus one loop for every baseline outside it — and adds up each loop’s vectors. For a perfectly consistent set of baselines every loop sums to zero, so whatever is left over is error.

Column	Meaning
Loop	The points traversed, in order.
Perimeter	Total length of the loop.
Misclosure X / Y / Z	The three components of the sum, in metres.
Misclosure	The magnitude of that vector.
PPM	Misclosure as parts per million of the perimeter. This is the figure to judge by, because it is independent of loop size.

Only baselines that are both included and solved can close a loop. A loop containing a failed or excluded baseline is counted in the loops skipped label rather than quietly dropped from the list — so if you see few loops and a non-zero skip count, the missing checks are the ones through your failures.

Tip

A loop misclosure much larger than the rest is the signature of one bad baseline, not of general noise. Look for the baseline that all the bad loops have in common, and re-solve or exclude it.

2.3.5.3. Occupation Timeline

A time chart of the campaign: one horizontal bar per file, drawn on a common time axis, so you can see at a glance who was observing when, where the overlaps are, and where a session boundary falls.

Reference stations are drawn in a darker blue than the rovers.

This is the fastest way to understand why the session grouping came out the way it did — a rover whose bar barely touches the base’s bar is the one the warnings panel is complaining about.

2.3.5.4. Network Sketch

A plan view of the points, positioned from the approximate coordinates in the RINEX headers, with the baselines drawn between them. It is a topology check, not a survey plot.

Baselines are coloured by outcome, so the sketch shows you where in the network the trouble is rather than just that there is some:

green — solved, FIX;
orange — solved, FLOAT;
red — failed.

Only included baselines that have been attempted are drawn; a baseline still in the planned state does not appear.

Use the mouse wheel to zoom, drag to pan, and double-click to fit the whole network back into the panel.

Note

One wrong header can put a point thousands of kilometres from the rest. Rather than let that collapse every real point into a single dot, the sketch excludes points whose approximate position is a statistical outlier against the rest of the network, and says so in a label under the drawing.

If a point you expect is missing from the sketch and named in that label, its RINEX APPROX POSITION XYZ is wrong. That does not necessarily hurt the solve — the solver refines the position — but it is worth knowing about.