| 22 Jul 2026
Quantifying the effects of observational time resolution on marine heatwave detection
Abstract. Marine heatwaves (MHWs) are increasingly recognized for their significant impacts on coastal marine ecosystems. Following a proposal by Hobday et al. (2016) MHW are often defined as an event when the sea surface temperature (SST) is larger than the 90th percentile of the local climatology, during at least 5 consecutive days. The definition of MHWs presents a methodological challenge: at high temporal resolution, short-term excursions below the 90th percentile threshold can interrupt the continuity of an event and thus invalidate it as a MHW. In contrast, when the data are averaged over longer timescales (e.g., 10 hours), such fluctuations are smoothed out, potentially allowing the same event to meet the MHW criteria. A relaxed version (allowing 2-day gap) of the definition consists in discarding short-time interruptions. We tested the influence of temporal resolution by calculating the number of marine heatwaves detected in 6 coastal stations situated along the French coast, across different temporal aggregations, from raw (e.g., 20-minute) resolution to daily averages. Because short timescale variability generates numerous brief threshold exceedances, finer resolutions yield more frequent but shorter events, whereas coarser resolutions reduce frequency, maximum intensity and inflate duration. Using these coastal stations, we show that marine heatwave (MHW) metrics are systematically sensitive to changes in temporal resolution and event detection criteria, with strict and relaxed definitions converging—and in some cases crossing—as data are aggregated to coarser timescales. Overall, the resolution dependence of MHW metrics is systematic across sites, underscoring the need to consider the time scale when estimating and comparing MHWs.
General comments
The manuscript proposed by H. Chavan et al. discusses the impact of observational time resolution on marine heatwave detection.
It discusses the general methodology for detecting and characterising marine heatwaves in relation to observation time resolution. In light of the growing number of high-resolution observation systems, the manuscript cautions against the use of a general definition for high-frequency observations.
The manuscript is highly focused, exploring limitations based on in situ observations in contrasting coastal regions.
As a general comment, I suggest that the manuscript would benefit from more discussion in comparison with the extended literature on marine heatwaves. For example, the results could be discussed in relation to Figure 4 from the 2016 paper by Hobday et al., where in situ observations show less variability than daily Sea Surface Temperature observations from satellites, which seem to differ from the author's observations.
Considering the limited improvements needed, as detailed in the general and specific comments, I recommend this paper for publication after minor revisions.
Specific comments
l. 112, p. 4 – The MesuRho station is located in front of the Rhône River delta, rather than inside it.
l. 107–116, p. 4 - The sensor is not specified for the MOLIT station (most probably also the NKE probe), but I guess it should be a similar type of probe to those used for the other stations ("MPx") instead of Smatch, as it is an older system.
Section 3.1 – How are probable existing gaps in the data considered in the climatology computation? Do they have an impact on the climatological values? Do they explain part of the smoothed noise?
l. 149, p. 5 – Is the resampling method based on averaging or subsampling at the target resolution?
Section 3.3 – Regarding the methodology and the strict and relaxed detections, gaps in the in situ data can also affect the detection, as they can virtually disrupt the time series. Were the gaps filled before the MHW analyses? If not, did the authors evaluate the impact of the gaps on the results?
Section 4.1 - A strong and moderate classification is used for MHW, but this is not defined in the manuscript. What intensity classifications are used in the manuscript, and how are they related to the classification used by Hobday et al. (2018) (https://doi.org/10.5670/oceanog.2018.205)?
l. 255–256, p. 9 - The authors mention stations dominated by short-lived spikes. It would be important to explain why the spike dynamics are different for these stations compared to others, given that they are in very different locations.
l. 260–265, p. 9 - For the Mesurho station, peak intensities are explained in terms of wind dynamics, but the influence of the Rhône river plume at the station is not mentioned. Does the river plume not play a role in peak intensities?