2.3.8. Worked Example

A complete pass through a small static campaign: two base stations running all day, five rovers occupying marks in two sessions.

2.3.8.1. Before You Start

Put everything in one folder:

the observation files, one per receiver per occupation;
the broadcast navigation files covering the same days.

The scan is not recursive, so a campaign split across day1\, day2\ subfolders needs one scan per folder — or flatten it first.

2.3.8.2. Step 1 - Open a Project

Calculations > GNSS Processing, then New Project…. Give it a name and a folder; the file gets the extension .adjproj.

This is an ordinary adjustment project — the same file the Network Adjustment module opens — which is why the baselines will not need exporting later.

2.3.8.3. Step 2 - Scan and Name

On the Files page, press Scan Folder… and pick the campaign folder. The grid fills with one row per observation file.

Now do the two things only you can do:

Fix the point names. The MARKER NAME column shows what the receiver recorded; the Point Name column is what your project calls the mark. Correct every one that is wrong. Two files given the same point name are treated as the same mark — which is right for a re-occupied base, and wrong for two different marks that happen to share a receiver ID.
Tick the references. Mark the two base stations.

Then read the Recognised, not used list under the grid. It names every file in the folder that did not become an observation or navigation row. Field notes and archives listed there are harmless; a navigation file listed there is not — it means the file was refused as ephemeris, usually because it is an almanac or an SBAS file rather than broadcast navigation data.

2.3.8.4. Step 3 - Sessions

Go to Sessions & Baselines and press Rebuild sessions.

The tree should show the sessions you expect. Read the Warnings panel before going further:

a session with one file in it, or a file excluded from every session, means the Min overlap (s) value is not right for this campaign — the default is 900 s;
a file with no TIME OF LAST OBS needs re-converting before it can be used at all.

Fix what the panel reports, change the minimum overlap if you need to, and rebuild until the grouping matches what actually happened in the field.

2.3.8.5. Step 4 - Plan the Baselines

Choose a Strategy and press Re-plan baselines.

For a campaign with real base stations, Rover x references is the right choice: it ties every rover to both bases and the bases to each other, which gives the network something to check against.

Read the last line of the warnings panel:

18 baselines · 7 points · redundancy 11

If the panel is red and reports redundancy zero, stop and fix it — a zero-redundancy network cannot be checked, no matter how good the individual baselines look.

Everything the planner produced starts ticked. Untick anything you know is not worth processing — a pair with two minutes of overlap, say — and watch the redundancy line as you do.

2.3.8.6. Step 5 - Process

The page already opens on this module’s defaults, which suit a static control campaign:

Mode: Static
Constellations: whatever your receivers tracked
Elevation mask: 15 degrees at an open site, higher where there are obstructions
Ionosphere: Off for short baselines, Estimate for long ones
Troposphere: Saastamoinen
Ambiguity resolution: Fix-and-hold (default)
GLONASS AR: Fix-and-hold (default) — but set it to Off when the two ends are different receiver makes
Ratio threshold: 2.0 (default)
Interval override: 15 or 30 s if the files are 1 s — much faster, and for static work costs nothing

Two of those defaults are deliberately permissive, so check the achieved ratio on the Results page rather than taking a fix on trust: fix-and-hold holds a wrong integer solution as firmly as a right one, and a 2.0 threshold accepts fixes RTKLIB’s own 3.0 would reject.

Press Run. The batch solves every included baseline that has not been solved yet, on a background thread, with a progress bar and a log. You can Cancel at any point without losing what has already completed.

2.3.8.7. Step 6 - Judge the Results

On the Results page, work through in this order:

The colours. Aim for all green. Orange is FLOAT, red is a failure.
The Notes column on every red row. It says why the solve failed.
The Ratio column on the green rows. A fix that just scraped over the threshold deserves a second look.
The loop closures. This is the real check. Judge by PPM rather than by the raw misclosure, because PPM is independent of loop size.
The loops-skipped count. If it is high, your checks are going through baselines that failed.

For anything unsatisfactory: use the Quality page on both ends of the baseline (see The Quality Page), change the relevant option on the Processing page, then come back, select the row and press Re-solve selected.

Repeat until you are happy. It is normal to go round this loop several times.

2.3.8.8. Step 7 - Control and the Free Check

On Control & Publish:

Tick Control for the points whose coordinates you know, and enter them — or press Import from project if the project already holds them.
Set the Constraint for each: 3D, 2D or Height only.
Set Datum to Free check (hold one point).
Leave Allow FLOAT baselines unticked.
Press Publish, read the preview, confirm.
Press Apply datum.
Press Open Network Adjustment and adjust.

This first run tells you whether your baselines agree with each other. Because only one point is held, your control coordinates cannot distort it. If the free check is bad, the problem is in the GNSS data and no amount of control will fix it — go back to Step 6.

2.3.8.9. Step 8 - The Constrained Run

Once the free check is clean, come back, set Datum to Constrained (hold all control) and press Apply datum again. Re-adjust.

Now any disagreement between your control coordinates and your GNSS network shows up in the residuals. A free check that was clean and a constrained run that is not means the problem is in the control, not in the GNSS — which is precisely the distinction the two-stage process exists to make.

Note

You do not need to publish again between the two runs. Apply datum touches stations only and never observations, so switching modes cannot duplicate or drop anything you have already published.

2.3.8.10. Checklist

Step	Check before moving on
Files	Every point name correct; references ticked; navigation files in the folder.
Sessions	Warnings panel clear of session and overlap problems.
Plan	Redundancy greater than zero; panel not red.
Process	Every baseline green (FIX), or a deliberate decision about each one that is not.
Results	Loop closures acceptable in PPM; few or no loops skipped.
Free check	Adjustment clean with one point held.
Constrained	Adjustment clean with all control held; no orphaned points reported by the publish.