Honing in on bioluminescent milky seas from space

We processed DNB imagery from three key regions per the historical record of mariner sightings²—the northwest Indian Ocean (5° S–20° N, 40–70° E) and Indonesian waters surrounding Java (15° S–0°, 100–115° E) and the Banda Sea (11–1° S, 120–135° E). Our search window spanned 2012–2021, during the periods December–March and July–September corresponding to the peak modes of ship sightings. Data processing and detection criteria are described in “Methods” section.

Our search yielded 12 DNB-detected events (listed in Table 1) whose properties met the strict criteria for milky seas. Physically unexplainable in terms of thermal emissions (which would require scene temperatures exceeding 600 K), uncorrelated with clouds/airglow, invisible during the day, and persistent over multiple consecutive nights, these luminous bodies drifted and evolved in ways that were consistent with the analyzed ocean surface currents. The start and end dates of detection were in many cases bound by the observable periods as defined by the lunar cycle. Here, we highlight three exemplary cases, with additional details for all cases summarized in Supplementary Discussion 2.

Socotra, July/August 2013

On 31 July 2013, the Suomi NPP DNB detected a luminous body with well-defined boundaries (Fig. 2), located east of Socotra in the northwest Indian Ocean, at (14.0° N, 57.0° E). Uncorrelated with the observed cloud field (Fig. 2a–c), the body drifted northeast with the currents at ~ 0.44 m s⁻¹, stretching and curving in a manner consistent with the analysed ocean-surface currents (Fig. 2d–f), which showed a clockwise-rotating eddy located to its south.

Whereas DNB imagery showed only dark ocean during the daytime overpasses of the same location, these glowing waters persisted on successive nights over a two-week period. By 2 August the milky sea covered ~ 9000 km² (involving roughly 5 × 10²¹ to 5 × 10²² luminous bacteria, per “Methods” section). The DNB lost sight of the milky sea on 14 August due to moonlight, and it was not seen again in the following moon-free period.

Suomi NPP’s daytime chlorophyll-a (Chla) retrievals, a proxy for the amount of organic material in the surface waters, showed structural similarities to the milky sea, but were more widespread (Supplementary Fig. S1). Moreover, the most elevated regions of Chla (> 1 mg m⁻³) occurred not directly atop, but adjacent to the most luminous waters—a recurring property among the cases documented in this research which may indicate regions of algal stress where (potentially luminous) bacteria would proliferate. On several nights, a faint signature of the luminous body was detectable beneath analyzed cloud cover; its light scattering upward through the clouds in a way similar to the behaviour of city lights in DNB imagery.

Somali Sea, January 2018

On 12 January 2018, both Suomi NPP and NOAA-20 captured a luminous structure offshore of southern Somalia. Over the next 5 days it stretched into a narrow filament that paralleled the Somali coast, mirroring the behaviour of other winter-mode Somali Sea cases described in Supplementary Discussion 2. Over 18–23 January, the luminous filament extended east/northeast, forming a comma-shape (Fig. 3a–c), with a sharply-defined southeastern edge and gradually fading brightness on its northwestern side. By 20 January, it spanned ~ 15,000 km², suggesting involvement of roughly 8 × 10²¹ to 8 × 10²² bacteria. Its scale, shape, time, and location were similar to the Lima-sighted milky sea^4,14, as well as to a subset of the surface reports in Supplementary Discussion 1.

Comparing the luminous body to satellite retrievals of Sea Surface Temperature (SST; Fig. 3d) showed its eastern boundary aligned with the edge of an oceanic front, residing within relatively cool (298–299 K) waters that extended northeast from the Somali coastal upwelling zone. These cooler SSTs corresponded to elevated Chla values in the range of ~ 0.5–1.0 mg m⁻³ (Fig. 3e). As in the 2013 Socotra case, the area of elevated Chla was more extensive than the luminous region. Ocean surface currents (Fig. 3f) showed the body’s eastern boundary embedded within counter-clockwise flow and drifting north/northwest at ~ 0.8 m s⁻¹.

Java, July–September 2019

The DNB detected a large milky sea in the east Indian Ocean, immediately south of Java, Indonesia in 2019. The event spanned two complete moon-free cycles (26 July–9 August, and 25 August–7 September). On the night of 25 July, the DNB detected a luminous anomaly south of Surakarta, Java, near 9.5° S, 111° E. The detection amidst moderate moonlight conditions suggested a particularly strong source of emission. Imagery on subsequent moonless nights confirmed that the initial detection was in fact part of a much larger milky sea, spanning ~ 100,000 km²—approximately the same size as Iceland. A milky sea of this scale suggests involvement of roughly 6 × 10²² to 6 × 10²³ luminous bacteria, which would qualify as the largest event on record. Undetectable during the day, the contiguous feature reappeared in nightly imagery throughout the two observable moon-free periods (Fig. 4).

Situated within quiescent, low-shear waters (Supplementary Fig. S2) between counter-clockwise-spinning warm-core eddies to its southeast and southwest (Fig. 4), this massive milky sea rotated clockwise like a cog between gears, its centre near 9.0° S, 110.0° E. By 30 July, its northern boundary approached within 25 km of the Java coast, and a 500 km² area of its core was so bright that certain infrared-detected cumulus clouds appeared in the DNB imagery as dark, attenuating objects in contrast to the glowing waters below them (Supplementary Fig. S3).

The DNB radiances measured in the brightest areas of these luminous waters approached Crescent- to Quarter-Moon illumination levels. Based on scotopic vision sensitivity to bioluminescent emission (Supplementary Fig. S4) and direct comparisons against legacy OLS imagery (Supplementary Fig. S5), portions of this milky sea may have appeared visually bright to dark-adapted human vision—perhaps even attaining the classical snowfield effect described in the historical mariner accounts.

After losing sight of the luminous body on 10 August due to moonlight contamination, the DNB recaptured it on 25 August and tracked it for 2 weeks thereafter, as described in Supplementary Discussion 2. The longevity of this event, which lasted for at least 45 nights, by far eclipsed all other cases encountered in this study—indicating that significant milky seas offer a reasonable time window for reaching them if a rapid-response team is on the ready.

Satellite retrievals of ocean surface properties for the 2019 Java case (Fig. 5) showed cooler waters and elevated Chla along the Java coast. A stream of higher Chla (> 2 mg m⁻³), embedded within a tongue of these cooler waters, extended southward and immediately east of the luminous body, following the flow of the south-eastern eddy (Fig. 5b). At this time, the luminous waters were confined to a narrow range of SST over 298 ± 1 K (~ 25 ± 1 °C) and moderate Chla over 1 ± 0.5 mg m⁻³, demarcated from the surrounding waters at abrupt boundaries defined by coastal upwelling to the north and the two eddies to the south.

Figure 6 relates DNB radiance, SST, and Chla for 10 nights (27 July–5 August) of the 2019 Java case, centred on the luminous body. DNB radiances correlated positively with Chla over 0.5–1.5 mg m⁻³, and negatively with SST over 297–299 K. The brightest milky sea waters corresponded to asymptotic SST values of ~ 298 K and Chla of ~ 1.2 mg m⁻³, respectively, and notably, did not overlap with the strongest parts of the algal bloom. The SST values, hovering around 298 K for most DNB-detected cases in this study, may hold significance, as this temperature regime promotes rapid growth of V. harveyi and P. leiognathi over a wide range of ocean salinity values²⁰.

Source: Ecology - nature.com