The manuscript concludes that the Mediterranean heat uptake, especially in the Aegean basin, can be used as a precursor to predict winter land precipitation in the Levant (eastern Mediterranean). The topic is interesting and the manuscript has the potential to be a useful contribution. But I think a few points need to be addressed before its publication. 

Firstly, many technical details are missing, which is harmful for a good understanding of the manuscript. For example, it is really difficult to figure out what presented in Figures 2b, 2c, 3b and 3c (tables showing correlation coefficients). 

Secondly, the use of SOM is unusual and not justified in this work. Normally SOM is used with a 2-D grid of different nodes, but there are only three patterns in the manuscript. 

Finally, the manuscript concludes that AQA (Aegean Sea heat uptake anomaly) is a good indicator for precipitation anomalies in the Levant. But, compared to SST which is a state variable, the heat flux is much more difficult to measure or to be deduced from observation. Its usefulness might be quite limited. Furthermore, it is also a little disappointing that a clear physical mechanism is missing to link the AQA to the precipitation in the target area. 

There are a few other points:

1. Figure 1 is used to motivate the present work exploring the role of the Mediterranean Sea in modulating precipitation in the Levant. But it is not very convincing. The signal is not remarkable in the Mediterranean, but much stronger in other basins of the global ocean.

2. Line 106, “The SOM algorithm is applied to detrended monthly deviations from the climatological seasonal cycle”. It is not clear how SOM is performed. Is it applied to anomalous SST, i.e., SSTa(m7:m11,y1979:y2023)? Is there any coherence or consistency among m7 (July) to m11 (Nov) for a same year?

3. Figs 2 and 3, Figure Caption, “SST monthly time series”. There is confusion for the term “time series”. More precisions are needed.

4. Panels 2b, 2c, 3b, 3c. How is calculated the temporal correlation? between what and what? 

5. Line 207, “AQA is strongly correlated with Qf Pattern 2”. AQA is a time series, but the Qf SOM pattern is a geographic structure. How can they be correlated?

This paper focuses on an important challenge of improving seasonal prediction of winter precipitation in the Levant region by connecting it with Mediterranean Sea heat uptake patterns. The authors use Self-Organizing Maps (SOM) to identify three main spatiotemporal patterns and develop the Aegean Qf Anomaly (AQA) index representing SOM 2, which correlates strongly with Levant winter precipitation. The research provides valuable insights for seasonal prediction. I also liked that they backed their research with physical interpretation through hydrological decomposition and synoptic analysis.

While the overall approach is solid, I have several concerns about methodological choices and explanations that need addressing. Here are my main points:

Major comments

1- SOM vs. EOF choice:

-The authors mention SOM's advantage of not requiring orthogonality, but don't fully explain why this matters for this specific analysis.

-If EOF produces similar patterns as, including at least one EOF figure in the Supplement would strengthen this justification.

2- SOM parameter selection:

-The optimization approach - maximizing correlation with precipitation while minimizing pattern count - seems somewhat methodologically questionable. Why is maximizing correlation with the target variable appropriate for what should be a self organising clustering technique?

3- AQA index definition

-The fixed Aegean Sea box used for the AQA index appears to be visually selected based on SOM Patterns 2 and 3. While it is an effective choice, I recommend discussing the robustness of the AQA box definition.

4- Correlation analysis details

-The key correlation between August AQA and winter precipitation (R = –0.60) needs more context. Please clarify exactly how Levant precipitation is calculated (sum/average over the region) and include confidence intervals for the correlations in Fig. 5b.

5- Physical mechanisms

-While the composite analysis showing Cyprus Low persistence and subtropical jet strengthening is convincing, I would be cautious claiming independence from remote drivers (NAO, ENSO, etc.). Lack of correlation doesn't necessarily mean lack of physical connection. A mechanistic explanation would be more persuasive.

Minor comments

- In abstract be specific that R = –0.60 refers to the correlation between August AQA and DJF Levant precipitation.

- In methods when describing heat uptake (Qf), clearly state that positive values mean the ocean gains heat and negative values mean heat transfer to the atmosphere.

- Figures 2–3: Include the cumulative variance explained by the three SOM patterns.

- Figure 5: Add statistical significance indicators or confidence bounds to the correlation values.

- Vague phrases like "optimal results" should be clarified - do you mean statistical robustness, strongest correlation, or something else?

Please see attached file