Scene-level validity of the compact hyperspectral retrieval

Determine whether the compact symbolic-regression chlorophyll-a expression can be applied to spatial satellite imagery without introducing discontinuities or unrealistic spatial structures.

Background

The paper evaluates the compact PACE-like symbolic-regression expression on held-out matchup observations and characterizes its performance across chlorophyll and SST regimes. However, matchup-based diagnostics do not determine whether the expression behaves realistically when evaluated pixel by pixel over complete satellite scenes.

The authors therefore identify scene-level spatial behavior as an unresolved validation issue, specifically whether applying the expression to imagery could produce discontinuities or physically implausible spatial patterns. The subsequent scene examples demonstrate feasibility but lack collocated in situ observations for definitive validation.

References

The regime analysis characterizes where the models perform well or poorly within held out matchup observations, but it does not establish whether the compact expression can be applied to spatial satellite imagery without introducing discontinuities or unrealistic structures.

— How Much Hyperspectral Information Does Chlorophyll Retrieval Really Need?  (2609.18531 - Hammoud et al., 16 Sep 2026) in Section 4.6, “Regime diagnostics”