Calibrate and connect the record

A detector produces an electrical response recorded as digital counts. Calibration connects that response to a physical measurement scale. NSMC describes preflight calibration and later checks using reference targets, comparisons between instruments and field observations. Those later checks matter because a sensor’s response can change in orbit.

Consider an explicitly simplified exercise: a stable reference corresponds to 250 K, but a product repeatedly reports 252 K under matching conditions. The observed offset is plus two kelvin. If that constant-offset model has been validated over the relevant range, a reported 260 K would correct to 258 K. One reference alone does not establish that the same correction works at all temperatures or viewing conditions.

Feng Yun 3E carries X-EUVI, which observes the Sun in X-rays and extreme ultraviolet for space-weather work. The selected figure shows a 19.5-nanometer solar observation on November 5, 2021. Its orange color is a display choice for radiation invisible to human eyes; the figure retains its physical scale and observation time.

The instrument paper describes a companion irradiance sensor for radiometric calibration and repeated rotated solar images for flat-field correction. Flat-field work addresses differences in response across the detector. Radiometric calibration establishes the physical measurement scale. These corrections answer different questions, and agreement with another instrument is useful evidence rather than proof of perfection.

When sensors are replaced, a jump in a record might reflect a real environmental change, a different observation time, an instrument difference or processing changes. Overlap measurements, common references and documented processing help separate those possibilities. A long time series should retain instrument identities and version information.

In a 2023 conference abstract, NSMC authors describe rescuing and reprocessing historical Fengyun observations, including records beginning in 1988. Reprocessing can improve consistency; it does not mean every uncertainty disappears. Keep the original observations and explain the revised calibration and product version so that a later user can reproduce the analysis and understand what changed.