For each event, the authors describe the atmospheric-correction chains, the spectral indices used to flag bloom waters, and the spatial-resolution experiments applied to each product. The result is a reproducible cross-sensor comparison workflow and an explicit sensor-selection rationale, rather than a single-number verdict on which satellite is best.

The 30 April event revealed how sharply sensor choice can change the apparent size of a bloom. Under an acquisition difference of roughly ten minutes, GOCI-II delineated a bloom area of 901.52 square kilometers, while Sentinel-2B mapped 571.52 square kilometers of the same event. That gap of more than 300 square kilometers does not mean one satellite was simply wrong. According to the authors, the contrast reflects the joint effects of spatial resolution, spectral configuration, atmospheric correction and threshold choice, and the observations cannot isolate any single factor as the sole cause. It is a crucial caution for anyone estimating bloom extent from a single satellite product.

The September event showcased the unique strength of geostationary observation. Because GOCI-II stares continuously from its orbital position, it collected hourly images throughout the day, resolving an intra-day sequence in which the mapped bloom area surged from 37.30 to 760.20 square kilometers and then fell back to 107.67 square kilometers. The bloom’s centroid, its geometric center of mass, displaced by 45.74 kilometers over the same period. Single overpasses from Sentinel-2 and Sentinel-3, which sweep across the region only once per pass, sampled only part of the event.

This temporal dimension matters because, as the September sequence itself demonstrates, bloom extent in optically complex coastal waters can change dramatically within hours. GOCI-II’s hourly cadence effectively turns the satellite into a bloom movie camera, making it the natural choice for temporal screening and early-warning applications, even though its spatial detail cannot match the finer resolution of Sentinel-2.

To probe how spatial resolution itself shapes bloom mapping, the researchers ran a degradation experiment in the surface-reflectance domain, deliberately coarsening the imagery to mimic coarser sensors. The results were striking and asymmetric. For a fragmented, patchy bloom, the measured boundary length declined by 76.2 to 85.4 percent after degradation, while for a compact patch the decline was 80.3 to 82.7 percent. Area, by contrast, changed far less dramatically, ranging from a 14.8 percent decrease to an 11.0 percent increase depending on the case. In other words, coarse resolution preserves the overall footprint of a bloom far better than it preserves the boundary complexity that matters for tracking how a bloom disperses.

These findings converge on a clear division of labor among the three sensors. GOCI-II excels at temporal screening, catching the onset, peak and decay of blooms that evolve within a single day. Sentinel-2, with its high spatial resolution, is suited to fine-scale mapping, delineating individual patches and their boundaries with precision that coarser sensors cannot achieve. Sentinel-3 OLCI contributes regional spectral context, its ocean-color heritage making it well suited to characterizing bloom conditions across broad areas. No single platform dominates; the optimal monitoring strategy is complementary.

The practical implications extend to operational harmful algal bloom early-warning and nowcasting in optically complex East Asian coastal waters. By publishing the atmospheric-correction chains, index choices and threshold decisions for each sensor, Zhang and Shi have provided a transparent, reproducible workflow rather than a single-number answer. Analysts can now justify sensor selection for a given monitoring task instead of defaulting to whichever satellite happens to pass overhead.

The study also carries a methodological warning for the broader remote-sensing community. Because the observed area discrepancies between sensors arise from the interaction of resolution, spectral bands, atmospheric correction and threshold selection, comparisons of bloom statistics across different satellites should be interpreted with care. A bloom that appears to grow or shrink between two products may simply reflect how each instrument and processing chain samples the same water. The degradation experiment quantifies this effect directly, showing that boundary complexity is far more sensitive to resolution than total area.

For coastal managers in the East China Sea region, where harmful algal blooms threaten aquaculture and nearshore environmental security, the message is that a multi-sensor strategy is a necessity rather than a luxury. Hourly geostationary data can screen for blooms as they develop, high-resolution imagery can map the affected patches in detail, and regional ocean-color observations can place each event in its broader spectral context. The authors’ framework turns three satellites with very different strengths into a single, coordinated observing system for bloom monitoring and early warning.

Subject of Research: Cross-sensor satellite comparison of harmful algal bloom monitoring in the East China Sea

Article Title: Multi-source satellite remote sensing of harmful algal blooms in the East China Sea: cross-sensor comparison of GOCI, Sentinel-2 and Sentinel-3

Article References: Zhang, C., & Shi, H. (2026). Multi-source satellite remote sensing of harmful algal blooms in the East China Sea: cross-sensor comparison of GOCI, Sentinel-2 and Sentinel-3. Frontiers in Marine Science, 13, Article 1966777. https://doi.org/10.3389/fmars.2026.1966777

Image Credits: AI Generated

DOI: 10.3389/fmars.2026.1966777

Keywords: harmful algal blooms, East China Sea, GOCI-II, Sentinel-2 MSI, Sentinel-3 OLCI, satellite remote sensing, ocean color, red tide, cross-sensor comparison, spatial resolution, coastal monitoring, early warning