Please see attached file for comments

This short paper investigates the predictability of the Marine Heatwave that occurred in the Mediterranean Sea in summer 2022. Using short-term (10 days) forecasts from Copernicus Marine Services, it examines if this system is able to successfully forecast the onset, evolution, peaks of this extreme event, first at basin scale, and then at two specific locations of economic importance. The main message is that short-term forecasts exhibit some skill in predicting details of the MHW evolution a few days in advance, which is of importance for guiding management.

This paper addresses an important challenge. Marine Heatwaves have recently received much attention because of their damaging impacts on marine ecosystems; several events have struck the Mediterranean Sea in recent years. It is of great importance to investigate their potential predictability at different lead times. While several papers focused on seasonal forecasts, this study focus on short-term (10 days), which has received little attention so far.

The paper is well written; the approach (first at basin scale, then at local scale) seems relevant to the objectives, and the results are worth being published.

Yet, I have two major concerns:

• the analyses presented are mostly qualitative, and would deserve more quantification. Sentences like “ the area of MHW conditions was also well predicted” (line 116) or “MHW occurrence […] correctly forecast” (line 137) or “upon visual inspection the forecast temperature was very similar to the observed” are subjective. What is a “good” forecast? How can it be defined? What are the acceptable errors? Spatial correlations, quantification of the differences, or other metrics to better quantify these are needed. Figure 4 is not fully discussed, yet the results do not seem very encouraging for 7-days lead time. Are the forecasts “good” for this lead-time?

• The analyses remain descriptive, without analyses on the processes that can lead to improved or degraded forecasting capability. Although I understand that a full characterization of the 2022 MHW drivers with a heat budget may be beyond the scope of this paper, it would be useful to provide more hints on the underlying processes, so that the scope of this study would not be restricted to this particular event but would provide information useful for other upcoming events.

In particular, one interesting result is the contrast between the two periods in early summer and late summer (May-mid-June versus July-August-September) as seen in Figure 1b. The operational system exhibit better forecasting skill during the early period, whereas it fails to predict the sharp increase in activity (intensity) in mid-July, early August, early September. Looking at the three specific periods to examine which processes in the real world led to MHW decrease instead of intensification, would help to understand these false alarms. Line 134, the authors suggest as potential candidates “cloud cover or winds”. Would it be possible to investigate this?

Finally, some details are missing. The area chosen for the definition of the Ligurian Sea and Gulf of Taranto (shown in Figure 3) are not given. In Figure 4, it should be stated that the Figure is the basin average.

Minor comments:

-line 24: McAdam et al. 2023 reference is missing

-line 34: I disagree with the statement that 5 days of longer are harmful to marine life. This is completely species-dependent. The Hobday et al. (2016) definition is just a practical one, without reference to ecosystems impacts.

- line 74: no trend is removed. This choice may be usefully justified, given the recent debates from the community about removing or not a trend for MHW detection