FORMOSAT-3/COSMIC: turning navigation signals into atmospheric profiles

FORMOSAT-3 and COSMIC are names for the same Taiwan–United States observing partnership, not two competing constellations. Six small satellites launched in April 2006. An individual member such as COSMIC 1-1 belongs to that larger system. UCAR records the mission’s retirement in 2020, while the observations remain available for later research. A spacecraft’s operational lifetime and a dataset’s useful lifetime are different things.

The partnership used radio occultation — open definition to study the atmosphere and ionosphere. Navigation satellites already transmit radio signals. A receiver on a lower spacecraft observes those signals as the transmitting satellite appears to rise or set behind Earth’s edge. Changing atmospheric conditions affect the signals along their path. The experiment uses that propagation information rather than taking an ordinary photograph of clouds.

As the geometry changes, successive signal paths probe different atmospheric levels. Analysts use the observations together with precise orbit and clock information to derive quantities describing the atmosphere. A vertical profile — open definition reports how a quantity varies with height. It is not a physical thermometer lowered along one perfectly vertical line, and the processing assumptions matter when comparing it with other measurements.

COSMIC observations supported weather prediction, climate studies and investigation of the ionosphere. Coverage over regions with few conventional observations made the measurements particularly useful. The result still has to be combined with other evidence and a model for a forecast. One profile does not describe every location, determine every storm track or eliminate uncertainty in the evolving atmosphere.

Observation time is not processing time

UCAR’s archive distinguishes tracking data, orbit and clock products, and derived atmospheric profiles. Its later reprocessing of the mission record applies a documented processing version to existing observations. A new product release is not evidence that a retired satellite launched again or collected fresh measurements. Observation time, processing version and publication time belong in separate fields.

Consider a fictional comparison in which an older retrieval — open definition gives 250 kelvin at a specified level and an updated retrieval of the same observation gives 251 kelvin. The one-kelvin difference cannot by itself establish that the atmosphere warmed between those releases: the underlying observation is unchanged. To investigate real change, compare suitably calibrated observations from different times using consistent methods. Preserved data and clear processing histories let researchers ask new questions without confusing an improved analysis with a new event.