Better coastal satellite images still need better measurements
WHAT THE EVIDENCE SAYS
A cross-sensor coastal mapping workflow used 18 same-day Sentinel-2–PlanetScope image pairs for radiometric calibration and generated 191 synthetic supervised pairs. The tested reconstruction aimed at five-meter detail from ten-meter Sentinel-2 inputs. In ablation testing, radiometric harmonization was the largest individual contributor to reconstruction quality at fixed training-set size.
WHY IT MATTERS / OUR READ
An attractive high-resolution image is not necessarily a more accurate habitat map. Cross-sensor agreement and field-validation targets should be designed before treating sharpened imagery as evidence of restored shorelines or wetlands.
CAVEATS / UNCERTAINTY
The model was calibrated to Galveston and must be recalibrated elsewhere. Neither downstream ecological mapping accuracy nor coastal restoration performance was demonstrated.
SOURCE
Dezhban & Koola · Frontiers in Marine Science · coastal image-harmonization study (16 Oct 2026)
View original evidence ↗
A Galveston super-resolution workflow highlights sensor harmonization; it has not demonstrated a biological outcome.
Read the Wild Lands assessment →