Under the most widely used definition, a marine heatwave starts when water temperature exceeds the 90th percentile of the values expected for that day of the year for at least five days in a row, and ends when it drops back below. Everything is measured inside that window: duration, peak, accumulated heat. Outside it, statistically, the heatwave does not exist.
Ricardo Nardi and Piero Mazzini, of the Virginia Institute of Marine Science, and two colleagues looked outside the window. They used temperatures from 54 stations in 20 estuaries of NOAA’s National Estuarine Research Reserve System, each with at least twenty years of data between 1995 and 2023. Of 3,097 heatwaves, 2,587 had complete records before and after the event. The work was published in Communications Earth & Environment on 10 June 2026; the press release that relaunches it is dated 30 September, almost four months later.

In their definition the phase before the heatwave starts on the first day the water rises above the average after at least three days below it. The phase after ends on the last day above the average before three days below. Accumulated heat is counted in degree-days: one degree above the average for one day equals one degree-day.
The surrounding phases last longer than the heatwave. The median heatwave lasts 7 days, the phase before it 11, the phase after it 9. They are less intense, about 1.2 degrees above the average against 2.91 during the heatwave, but time makes up for it. The median heatwave accumulates 20.6 degree-days, the phase before 11.8, the phase after 10. In about six events out of ten the heat outside the heatwave exceeds the heat inside it.

That is where the press release figure comes from. The authors add the median heatwave, 20.6 degree-days, to the median heat before and after, 31.6, and obtain a total 153 per cent higher. It is a comparison of medians, not an average. Taken event by event, the surrounding heat is a median 1.4 times that of the heatwave, but the spread is wide: in a quarter of events it is less than 0.6 times, in another quarter more than 2.9. The discussion also gives a third figure, 37 per cent. The direction is always the same; the size depends on how you count, and it does not hold for every event.
The real difference lies between two kinds of heatwave. Two in three are isolated: the anomaly starts about thirty days before and fades about twenty days after, and the warmth around it stops at a median 87 per cent of the heatwave’s. One in three, 843 events, is compound, with another heatwave inside the phase before or after. In these cases the water stays warm for up to ninety days on either side, and the surrounding heat is 3.6 times that of the heatwave.

The authors measure temperature, not damage. They say so: turning heat into effects on a species is a separate question that depends on the organism, its life stage and the season. They note, however, that warm water for longer means less dissolved oxygen, more acidification and conditions that favour toxic algal blooms. And they cite experiments on the California mussel, Mytilus californianus, in which slow, moderate warming induces acclimation and improves survival of a later heat extreme, while abrupt warming leads to cardiac collapse. The shape of the curve counts, not just the peak. Many laboratory experiments simulate a heatwave with a few days or weeks of heat, without the before and after.
The limitations are stated. NOAA’s stations sit mostly in shallow, well-mixed water, which responds differently from deep channels, the open coast and the open ocean. The series were detrended to remove long-term warming, a choice that, the authors themselves note, may understate the risk for organisms that adapt slowly.

The data cover only the United States. Italian lagoons and estuaries are shallow systems like the ones studied, but with different tides, salinity and circulation, so the result does not transfer as it stands. The method does: the code is public on GitHub and the summary data are deposited in the repository of the College of William & Mary.
On paper a heatwave has a beginning and an end. In the water it often starts earlier and finishes later.

References
Nardi R.U., Mazzini P.L.F., Walter R.K., Shen J., «Persistent warm water anomalies before and after marine heatwaves amplify heat exposure and associated risks», Communications Earth & Environment, 7, 1, 776, 10 June 2026, open access. doi: 10.1038/s43247-026-03739-x
Nardi R.U., Mazzini P.L.F., Walter R.K., Shen J., «A foundational summary dataset for “Persistent warm water anomalies before and after marine heatwaves amplify heat exposure and associated risks”», WMScholarWorks, 2026. doi: 10.21220/300
Virginia Institute of Marine Science, press release, republished by ScienceDaily on 30 September 2026 as «Ocean heatwaves are hiding months of extra warming».