Follow a tracer without losing the context

The NASA map in this lesson shows MOPITT-derived annual-average CO for 2014 at approximately 3,700 metres. Yellow through orange to red indicates increasing concentration; gray marks unavailable data. It is a processed data visualization, not a natural-colour view of smoke. Its original full-frame file has no numeric colour bar, so this lesson uses its colours qualitatively rather than assigning precise values by eye.

The source describes concentrations in parts per billion by volume, or ppbv. A fictional value of 100 ppbv corresponds to a CO fraction of 100 in a billion, or one in ten million. That ratio differs from a total column expressed as molecules per square centimetre. Always check whether a figure shows a fraction, an integrated amount, an emission rate or a difference between products.

An annual map combines many observations. It can reveal broad geographic structure while hiding seasonal peaks and short-lived events. A gray gap is not a measurement of zero, and a map several kilometres above the ground does not by itself establish exposure at the surface. The observation year belongs beside the picture so a historical result is not mistaken for current conditions.

CO can arise from incomplete combustion, including fires and fuel use, and from atmospheric chemical production. Once present, it moves with the air and is removed by chemical reactions. This makes it useful for studying transport, but it also means a high concentration is not automatically a direct measurement of the emissions immediately underneath it.

Consider an original transport exercise with a plume travelling at a constant 10 metres per second for six hours. Six hours is 21,600 seconds, giving a distance of 216 kilometres. A downwind observation can therefore be separated from its source even in this very simple model. Real winds change, air mixes, chemistry proceeds and plumes can move vertically.

Attribution needs additional evidence: wind fields, fire observations, emission inventories, chemical models and other measurements can test competing explanations. A visible pattern is a starting point for a question. It is not enough on its own to assign responsibility to a particular facility, country or event.

Missing is different from zero

Clouds can prevent usable infrared retrievals. The Version 9 processing revisions admitted more land scenes that older cloud-screening rules had rejected, including some heavily polluted scenes. The study found improved coverage, with substantial variation by region. This was a change in data processing and accepted samples, not a new launch that made Terra fly over those places more often.

In a fictional five-day dataset, accepted values are 80 and 120, while three days have no usable retrieval. The mean of the two accepted values is 100. Filling the gaps with zero would instead produce 40 and invent a decline. Even the valid-sample mean may not represent all five days if the missing days systematically differ from the observed ones.

When comparing maps, keep the product version, season, sampled region and quality rules consistent, and report coverage. Better sampling can change a map’s average without a corresponding change in the underlying emissions. That is why a striking image and a numerical trend both need the processing record behind them.