Data quality checks

10 of 376 operating units flagged.

Rule: A unit whose status is operating and which also has a stop year.

Reading a hit: A unit can have stopped and restarted, or the stop year may be a planned one, so a hit is a question for the source, not a mistake.

16 of 53 retired, mothballed and idle units flagged.

Rule: A unit whose status is retired, mothballed or idle and which has no stop year.

Reading a hit: Not reported is a gap, not an error: the year the unit stopped is simply not given.

1 of 47 units with both a start year and a stop year flagged.

Rule: A unit with both a start year and a stop year, where the stop year comes first.

Reading a hit: A unit can have been rebuilt or restarted under a new start year, so a hit is a question for the source, not a mistake.

3 of 89 units under construction flagged.

Rule: A unit under construction whose latest planned start year is earlier than 2025, the year of the terminals release.

Reading a hit: A plan that has slipped is normal; this lists the ones where the date on record has already gone by.

2 of 142 units reporting both years flagged.

Rule: A unit whose construction year is earlier than its proposal year.

Reading a hit: Possible when work began on an earlier phase and a later one was proposed afterwards.

11 of 376 operating units flagged.

Rule: A unit whose status is operating and which has no capacity in Mtpa.

Reading a hit: These units are counted as units but add nothing to any capacity total on this site.

6 of 813 terminals flagged.

Rule: A terminal with no latitude or longitude.

Reading a hit: These terminals have a page but do not appear on the map.

922 of 3,803 pipelines with a route and a reported length flagged.

Rule: A pipeline with a route on the map where the length GEM reports and the great-circle length of the route we hold differ by a factor of more than 1.5.

Reading a hit: Often the route on the map is one section of a longer project, or the reported length includes lines not drawn. This is why the site never derives cost per kilometre.

2,153 of 4,246 pipeline projects flagged.

Rule: Pipeline projects with the same name, ignoring case, but different GEM project ids.

Reading a hit: 424 names are shared. Most are regional distribution networks filed under one name by province; the pages tell them apart by route, year or project id.

0 of 1,145 vessels with an IMO number — none flagged.

Rule: An IMO number where the seventh digit does not equal the weighted sum of the first six, modulo 10.

Reading a hit: 75 of 1,220 vessels carry no IMO number at all; those are a gap, listed in the data coverage page, and are not counted here.

1 of 2,618 pipelines reporting a capacity flagged.

Rule: The pipeline with the largest reported capacity, or the largest two or three together, where it stands more than 10 times above the next one down in the whole tracker.

Reading a hit: The rule judges a figure against the tracker’s own list and needs no outside idea of what is plausible. The figure may be in a different unit in the source; this site leaves it as reported on the pipeline’s page and keeps it out of the capacity ranking.

1 of 3,086 pipelines reporting a diameter flagged.

Rule: A pipeline whose widest listed diameter, converted to millimetres from the unit GEM states, is above 1,500 mm (about 59 inches).

Reading a hit: The widest gas trunk lines in service are around 1,420 mm, so a larger figure is worth checking against the source. It may be a different unit, or the diameter of something other than the pipe.

Does a flagged record mean the tracker is wrong?

No. A flag means two fields GEM reports sit oddly together under a stated rule. There is often an ordinary explanation, for instance a unit that stopped and restarted, or a map route that draws only part of a longer project. Each check says how to read a hit.

Which releases are checked?

The terminals release of September 2025, the pipelines release of November 2025 and the carriers release of June 2026, the three this site currently shows. The page is rebuilt with each new release, so the counts change when the data does.

Does this site correct the data?

No. It shows what the tracker reports, and a gap or an oddity is left visible rather than filled in. Corrections belong at the source, in Global Energy Monitor’s tracker; we can say what a given release contains.

Why are terminal coordinates not checked against country outlines?

Because the outlines this site holds are coarse (1:110 million scale) and leave out islands and overseas territories such as Puerto Rico and the Canary Islands, so the check would flag correct entries. A check that produces false alarms is worse than none.