I'd like to know what is causing the effect on the water depicted on the following screen (sentinel tile id: S2B_MSIL1C_20190805T144739_N0208_R139_T20PPA_20190805T162922 located near Saint Lucia). You can see a darker area on water were light seems to be decomposed resulting in heap of pixels of different colors (green, red, blue). I've checked the data quality report but this effect was not mentioned.
What you are seeing is not an anomaly at all.
The bright areas are affected by specular reflections between the sun and the sensor.
In the dark area, something is changing the surface of the water, causing the it to not have the specular reflection. This could be a (natural) oil slick, or a break in the wind field, causing the capillary waves to behave differently.
On the illustration below, you can see an example of a different optical sensor showing similar features.
For more info on the linked image, see https://visibleearth.nasa.gov/view.php?id=36873
The "heap of pixels of different colors (green, red, blue)" is caused by the fact that the different channels of Sentinel-2 MSI are capturing the light reflected by the waves at slightly different times and at a slightly different angles.
Details of the sensor configuration that causes this are available in the MSI Instrument Technical Guide. Table 2 on the referenced page lists delays for each band, summarized here (listing offset in seconds relative to B02):
| B02 | B08 | B03 | B10 | B04 | B05 | B11 | B06 | B07 | B8a | B12 | B01 | B09 | |-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----| | 0 |0.264|0.527|0.851|1.005|1.269|1.468|1.525|1.790|2.055|2.085|2.314|2.586| seconds
Because your RGB image has channels roughly 0.5 s apart, very short reflections (such as those of a glittering ocean) will look like either red, green or blue pixels. Another cause of bright spots on waves could be foam, but that would likely last longer than 0.5 s.
To try and see the temporal component, I extracted separate bands using EO Browser for the same scene, and I made a gif that shows the time-lapse of the data in the same order as they are observed by the sensor. (I excluded bands 10, 1 and 9 because their poor resolution made the video too jumpy)