The following methodology applies to this dataset in addition to other datasets published in Edmunds et al. (2019).

Methodology:
This study was completed on the coral reefs of St. John, which have been the subjects of time-series analyses for 32 years. The measurements described herein originated from a schedule of instrument deployments initiated in 2014 to quantify variation in underwater physical environmental conditions, and ultimately, to facilitate testing for their role in driving changes in benthic community structure. As part of this schedule, rainfall was recorded throughout the year, and a light meter was placed in Great Lameshur Bay in August 2017, with the objective of leaving it immersed for 6–12 months. Three weeks later, the first of two Category 5 hurricanes impacted the island, with the second arriving 14 days later. The discovery in July 2018 that this meter had survived the storms, and had remained upright and functional, created the opportunity to describe underwater light during, and immediately after, two major storms.

Rainfall was recorded on the north shore of St. John at Windswept Beach (18° 21´ 20.95N, 64° 45´ 57.53W), where a 20.3 cm, Standard Rain Gauge (NOAA, National Weather Service) was mounted on a roof, 1.5 m above the ground. This rain gauge was ~ 6.7 km from the underwater light sensor, and was emptied and read on a daily basis.

Underwater light was recorded with a light meter (Compact LW, JFE Advantech Co., Ltd, Japan) fitted with a cosine-corrected sensor recording photosynthetically active radiation (PAR, 400-700 nm wavelength) as photosynthetic photon flux density (PPFD). The meter was equipped with a mechanical wiper that cleaned the sensor before every measurement, and it was mounted with the sensor at 19.1-m depth on the eastern side of Great Lameshur Bay (18° 18´ 37.04N, 64° 43´ 23.17W. The instrument was operated in burst mode during which 10 measurements were recorded every 180 minutes, with 30 seconds separating measurements within a burst. This sampling regime ensured that the battery would support a deployment of one year. The Compact LW meter is designed for oceanographic applications to 200-m depth, is fitted with a photodiode sensor, and has an accuracy of ± 4% (over 0–2,000 µmol photons m-2 s-1) and resolution of 0.1 µmol photons m-2 s-1. The sensors are calibrated by the manufacturer, with the calibration stable for at least 1 year. When the meter was deployed in August 2017, it had been used underwater for ~ 16 mo in previous deployments, and initial records of PPFD were similar to those previously recorded at the same depth and time of year in St. John, which suggested that the calibration had not appreciably drifted.

PPFD also was measured on the surface, using two cosine-corrected sensors (S-LIA-M003, Onset Computer Corporation) mounted ~ 4 m above sea level on the roof of the lab, ~ 0.875 km from the underwater sensor. The surface sensors were attached to weather stations (Micro Station Data Logger H21-002, Onset Computer Corporation) that recorded light every 5 minutes. The two sensors were calibrated by the manufacturers, and were operated in a paired mode to detect spurious records and sensor drift, and to guard against equipment malfunction.