Build a reproducible atmospheric record

SCISAT-1 was designed for a much shorter mission than the record it ultimately produced. CSA’s 2026–27 departmental plan describes continuing scientific delivery and support for atmospheric research. That dated statement is stronger evidence of ongoing use than an older page whose extension horizon stopped at 2024; it does not promise a particular future retirement date.

A long record can be reprocessed when calibration, spectroscopy or retrieval algorithms improve. The ACE team’s January 2025 announcement explains the change from version 5.2 to 5.3, including work on spurious profile spikes near 30 kilometres. A change in a processed value can therefore reflect a better analysis of an old observation rather than a new atmospheric event.

Keep the input observation identifiers, processing version, quality flags and selection rules with the result. If comparing years, use a consistent product definition or quantify the effect of a version change. Simply joining old and new files at a date boundary can introduce an artificial step.

A 2025 study by the ACE research team improved retrievals of HFC-23, a greenhouse gas. The publisher’s abstract and research summary describe reduced relative uncertainty and improved agreement with ground-based AGAGE measurements. The updated analysis addresses interfering absorption, including HFC-125, and instrumental artifacts.

The study reports an HFC-23 increase of about 1.03 parts per trillion per year over 2012–2024. That finding belongs to the stated interval and research retrieval. It is not a universal growth rate for every location and altitude, and it should not be silently substituted for values in a standard product with a different version.

The ACE team explicitly distinguishes its standard molecular dataset from additional research products. Preserve that distinction when citing results. A paper describing an upcoming retrieval can provide important evidence without establishing that every publicly available data file has already been replaced.

Keep the evidence attached to a trend

Independent ground observations are valuable, but they need a fair comparison with a satellite’s sampled air. Different altitude ranges, locations and times can produce genuine differences. Agreement after accounting for those differences strengthens confidence; matching two plotted lines without checking their definitions does not.

In an original trend exercise, a fitted increase of 0.9 parts per trillion per year over four years implies a fitted change of 3.6 parts per trillion. It does not supply the starting concentration or prove that each individual year changed by exactly 0.9. Uncertainty and variability belong with the trend.

A useful atmospheric result states the molecule, spatial and time coverage, units, processing version, uncertainty and comparison method. Those details let another person assess how the evidence bears on chemistry or policy, and where additional measurements are still needed.