1. What GLD Process is

GLD Process turns the raw recording of an aerial methane survey into a result: it reads the files the ground station wrote during the flight, finds the leaks, works out how much gas each one is releasing, puts them on a map and writes the report for the client.

It is the web version of the Windows processing software: same criterion, same numbers, same table. It runs in a browser, so there is nothing to install on the machine of whoever does the processing.

The screenshots in this manual come from a demonstration flight: the track and the leaks are invented, but the numbers are not made up — the application processed that recording with its own criterion, exactly as it does with a real one.

2. Signing in

Access is restricted and every operator has a personal account. It matters more here than elsewhere: what comes out of this application is a document that goes to the client, and it has to be clear who processed the data and who approved it.

The 👥 button at the top right opens the account panel, and ⏻ signs you out. If you have no account, an administrator creates one — see Managing users.

3. The window at a glance

The window has four areas. Along the top the toolbar: the menus, the working mode and the controls that move the time window. In the middle the charts. Down the right side the readout of the selected leak with the severity scale, the leak line, and at the bottom the block of action buttons. The centre of the top bar always says which recording is open and what the application is doing.

4. Opening a recording

File → Open… (leak detection) and choose the folder of the flight. The application looks through it, groups the files by recording and asks which ones to process.

The chooser lists one row per recording, grouped by day, with a badge saying which logger the data comes from. Tick the ones you want, or All. In Manual the button reads Open; in Auto it reads Open & process, because there it also does the processing and the saving.

5. The pipeline

Before processing, load the client's pipeline: Maps → Load pipeline (KML / KMZ / GPX). The application will not process without it, and the refusal is deliberate.

The laser beam covers about ±25 m around the pipe. Beyond that corridor it was not illuminating the client's asset, so methane found there cannot be attributed to them — it may be a farm, a landfill, a neighbouring plant. Without the corridor the report would hand the client someone else's gas.

Once loaded, the distance from the pipe appears in the readout as Pipe dist. and as the Pipe m column of the table. Remove pipeline takes it off again.

6. Parameters

Process opens the parameters first; File → Parameters… opens them on their own. A detection has to pass every threshold at once — one alone is not enough, because a noise spike can beat any single one of them.

Concentration mode

OGMP — robust max is the correct, higher estimate: robust maximum over a 3-point average, plume passage 200 m. Oleg — 1st peak reproduces the exact numbers of the historical processing: concentration at plume entry, 30 m grouping. Use OGMP unless you have to match an old report.

Alarm — the four thresholds

Roof G2 and the two switches

Noise Level, Spec, Step and Leak Mark are the instrument settings inherited from the Windows software; leave them as they are unless you know why you are changing them. Group (s) is the one worth knowing: samples above threshold closer together than this become a single leak instead of a row each.

Range Finder uses the rangefinder channel, Fast Int switches to fast integration. Both mirror the switches of the Windows dialog.

The four gates against false positives

Exclude off-pipe — always on, and it needs the pipeline. Detections beyond the corridor become grey rows: out of the report, never deleted.

Gate detr+V — the spectral confirmation. It accepts a candidate only if there is an absorption-line record within Pair (s) of it and that line resembles the methane reference. This is what separates gas from a bump in the signal.

Dip depth — shape is not enough: the dip has to be deep. Left adaptive you declare how permissive to be (False alarms %) and the threshold follows the session; measured on a real campaign, 0.2% is the knee below which the sites with repeated absorption stop decreasing. Fixed, you set the minimum signal-to-noise of the dip by hand.

V-gate — off by default. It discards a cluster only when it is both short (below Run len) and without correlation evidence (below Corr (V)).

7. Processing

On OK the processing starts and a bar over the methane chart says how far it has got and what it is doing — reading the files, correlation, GPS track, gate, rows. It is the real progress, not an animation, and the number on the right is the length of the recording in seconds.

When it finishes, the top bar says how many leaks were found and how many samples were above threshold, and the charts fill in.

8. The charts

Two charts are shown to begin with: Methane Concentration and Normalized Concentration. Graphs ON adds the other three — Signal Strength (logarithmic axis), External Illumination and Range Finder — and the button then reads Graphs OFF to go back to two. The three extra ones tell you whether the instrument was working properly, which is what you look at when a recording gives a result you do not expect.

The charts share the time axis: the crosshair and the window move together on all of them. To look at a leak closely, narrow the window with the arrows next to its width and step through the leaks — or drag across a chart to zoom into that stretch, and press ↺ to undo it. Double-clicking a chart resets its vertical scale, and the wheel changes it.

The coloured strip between the charts and the panel is the severity scale — 50, 100, 200, 500, 1000 ppm·m. Those same colours mark the leaks in the table and on the map, and the thresholds are drawn as horizontal lines on the methane chart.

9. The readout

The panel on the right shows the selected leak. The concentration is highlighted in the colour of its severity band, so the magnitude is readable without reading the number.

10. The leak line

The leak line is the absorption trace of the selected leak: in red the methane reference, in blue what the instrument measured. Where there is gas the blue follows the red down — that dip is the spectral confirmation, and it is what the detr+V gate measures.

The sample is some sixty times smaller than the reference, so without magnification it would look like a flat line: the + and − buttons, or the wheel, zoom the vertical axis around the dip. The factor in use is written in the corner.

11. Moving between the leaks

Leaks is how many were confirmed, Show which one is being shown; the ◀ and ▶ arrows step to the previous and next confirmed leak, taking the charts, the readout, the leak line and the map with them.

⊛ Add leak adds a leak by hand at the time of the chart cursor — for what the operator sees and the criterion missed; its coordinates come from the GPS track, never typed. ✕ Delete leak removes the selected one.

12. The Detection Table

One row per leak, with everything that will end up in the report. The tabs at the top are a view filter — All, Confirmed, Grey — and they change only what is shown: the rows always stay in the data.

13. Reviewing the result

The criterion proposes, the operator decides. Reviewing means going through the detections one by one — ◀ ▶ to move, a narrow time window to see the plume, the leak line to check the absorption — and then correcting what needs correcting.

Grey rows

A grey row is a candidate that passed the thresholds but not one of the gates — the spectral confirmation, or the distance from the pipe. It is never hidden and never deleted on its own: OGMP reporting requires that what was discarded stay visible, and the operator has to be able to see what the machine set aside and why.

Off-pipe in one click

At the bottom of the table, ✕ Off-pipe (n) greys out in one go all the detections farther from the pipeline than the metres in the box beside it. The removal is not automatic on purpose: a plume drifts with the wind and the digitised pipeline has its own error, so the decision stays with the operator.

One plume, one leak

When the aircraft crosses the same plume more than once, the same leak comes out as several rows. Σ Average merges into one those closer together than the Distance in metres: the concentration kept is the highest and the position is the average of the ones merged.

Printing the list

Print list produces a printable page of all the detections, and the same list can be saved as CSV for Excel. It is the sheet you take into a meeting, before the report exists.

14. How much gas

The ∿ g/h button at the bottom of the table works out the flow rate of every leak and fills the g/h column. It needs the wind: the application fetches it for the hour and the place of the flight, and the status bar says so. With no connection, type it in knots in the box beside the button.

The figure is an estimate, and it should be read as one: it depends on the wind, on the height of the passage and on the width of the plume. It is right for ranking leaks and for the order of magnitude in the report, not for billing a volume.

15. The map

Maps → Session map puts the flight on satellite imagery: the track, the pipeline, and one marker per leak in the colour of its severity — grey for the ones the gates did not confirm. Clicking a marker gives the concentration, the correlation, the flow rate and the distance from the pipe, and selects that leak in the rest of the application.

16. Saving

💾 Save stores the leaks of the recording as they are now — after the merges, the removals and the additions. Nothing else is saved: the sensor files are never modified.

It matters because the campaign report is built from the saved recordings: one that was processed but never saved will not be in it.

17. The report

📄 Report asks in four steps for what the document needs. At the top of every step you choose whether the report is for the current session or for the whole campaign from the saved recordings.

18. The detection photos

Each detection page of the report carries a ground photo taken from the survey video, next to the satellite map. Step 4 has one slot per confirmed detection, labelled with its time and concentration, and there are two ways to fill them.

From your own videos (the usual way)

📁 Grab from my videos asks for the flight folder — the one that contains video/ — and fills every empty slot on its own. The video stays on your computer: it is decoded in the browser and only the small photos are sent. For a campaign, pick the folder that holds all the days: the matching is on absolute UTC time, so one pass covers them all.

When it finishes it says how many slots it filled, and how many were matched on time only — those are the videos without telemetry beside them, where the position could not be cross-checked. Review those.

From the server (when the videos are already there)

Under Advanced: video on the server you type the path of the folder on the server's storage and press 🖥️ Server grab: there the extraction is done by the server. It only works if the videos really are on storage the server can read — not on a colleague's PC.

Whichever way you use, the grabbed frames only pre-fill the empty slots: 📷 Upload photo… replaces any of them by hand, and nothing is ever filled in silently. A photo of about 1200×800 and under 2 MB is the right size.

19. A whole campaign

Switch to Auto and open the folder of the survey: the application processes and saves every recording one after the other, and the campaign panel shows how it is going — one row per recording, with its state and how many leaks it found.

Review opens one recording in the normal window — charts, table, everything — so it can be gone through by hand; the edits apply to the campaign as they are made, and 💾 Save confirms them. The ⏹ Stop of the batch interrupts it at the next checkpoint.

From the same panel, Print all detections and CSV give the list of the whole campaign — numbered exactly as the report numbers it — and Report campaign merges every saved recording into a single PDF.

20. Site emissions (OGMP)

File → Site emissions… (OGMP) answers a different question: not “where are the leaks along the pipeline” but “how much does each of these stations emit”. You give it the list of sites — one line per site, with name and coordinates — and the details for the cover, then it processes the campaign.

Every site gets an answer, including the ones with nothing: not surveyed, zero, or emission. The review panel is organised by site, in the order of the report, and under each station only the detections that fall in its area — so what the operator approves is exactly what gets printed. False positives are marked and stay visible, greyed, with a way back.

21. Exporting the data

File → Export data… writes the track and the leaks in the formats the GIS people ask for. You choose what goes in — track and leaks, track only, leaks only — and the colour and thickness of the track line.

22. Looking at the raw channels

File → TDMS Viewer (G4)… lists the files of a G4 recording with their groups and channels, and for each channel the length, the type and the minimum, maximum and mean. It shows the channels the detection does not use as well, which is why you open it: to see what the instrument actually recorded. Every file can be converted to CSV or text.

23. Managing users

The 👥 button opens Account & users. Everyone changes their own password there; administrators also create accounts, reset a password, grant or revoke administrator rights, disable and delete.

24. If something does not add up