Separate water light from other signals

Ocean-colour analysis seeks information in light emerging from water. The satellite signal also depends on the atmosphere and viewing conditions. A laboratory camera test establishes part of the instrument response; it does not automatically remove atmospheric contributions from a later scene. CONAE’s science plan includes atmospheric correction and separate algorithms for quantities such as chlorophyll-a and turbidity.

The 2023 design describes a visible/near-infrared camera and a near-/shortwave-infrared camera. Its band table assigns visible channels to water-colour applications and several longer-wavelength channels to atmospheric correction, including turbid-water cases. A channel can therefore be useful even if it is not intended to become one of the familiar red, green and blue display colours.

Use an original simplified signal budget in arbitrary compatible units. The measured total is 100; an estimated non-water contribution is 90, leaving ten units assigned to the water signal. If that contribution was actually 92, the true residual would be eight. The two-unit correction error is only two percent of the total, but twenty-five percent of the true eight-unit water residual.

This subtraction model isolates sensitivity; real atmospheric correction is more complicated and the values are not SABIA-Mar observations. It shows why a visually plausible scene can still contain a scientifically important correction error. Calibration, atmospheric assumptions and quality flags need to travel with the resulting product rather than disappear behind its colour palette.

A small correction error can matter to the water signal

A spectral ratio compares measurements at selected wavelengths. In a fictional example, corrected reflectances of 0.02 and 0.04 produce a ratio of 0.5. If both share the same multiplicative factor, the ratio can cancel that particular factor. It will not generally cancel an additive offset or errors that differ between bands. An algorithm using a ratio still needs calibration and validation.

Chlorophyll-a is a pigment-related quantity, not a direct count of fish. Suspended particles and dissolved material can also affect water colour, especially near coasts. A derived map can help investigate habitat or productivity when interpreted with other evidence. It cannot, by itself, establish the number or species of fish, or prove that a vessel committed a legal violation.