In their article “Assessment of seasonal soil moisture forecasts over Central Mediterranean toward groundwater management”, the authors investigate the performance of SEAS5 soil moisture predictions across various lead times compared to ERA5 reanalysis, employing several performance metrics. The evaluation spans the period 2001-2021, focusing on the Central Mediterranean region. The findings reveal promising forecast accuracy for specific regions and soil layers, particularly at a depth of 289 cm.

General comments

Overall, the manuscript is well structured and well written. The topic of this research – the analysis of seasonal forecasts - is an area of relevance and interest. However, I have a couple of general comments: (1) There is a need for more detailed explanations and discussions on the performance metrics utilized in the study. Referencing other studies that employ similar metrics would enhance the overall quality of the article. (2) To fully support the conclusions of this study, more case studies are needed. In their case study, the authors did not account for the antecedent moisture content, which is an important factor in understanding the dynamics of the studied phenomenon. I would greatly benefit the overall quality of this article to analyse additional events, e.g. to include case studies that examine wet/dry events occurring after a dry season versus after a wet season, to provide a more comprehensive analysis of the system's predictive ability to varying moisture conditions. Alternatively, comparing SEAS5 to additional soil moisture products such as SMAP, ESA-CCI, or in situ observations if available, would also support the conclusions of this research.

Specific comments:

Introduction:

• This study is focussing particularly on SEAS5 performance. However, while ERA5 performance is stated and cited in the intro, the same is missing for SEAS5. Where do these forecast generally show best performance on a global scale? What are previous studies etc.
• What other climate services are available in general and specifically for this region? Are there already services available that are used by government/agriculture?

Methods:

• The metrics used here should be supported by more references and additional statements on their applicability, and interpretability.
• Why use the ROC metric and not plot the curve once?
• Are there no in situ observations of soil moisture available at all?

Results:

• Overall, the results section is well structured. Some statements belong to the discussion or conclusion.
• It would be interesting to plot the performance of the whole ensemble of forecasts in one plot and comment on the spread.

Conclusion:

• What is the relevance of this study, like development of real-time application for climate services as mentioned in Line 183 – 184. This is clearly stated in the title of but is not clear from the text. Mention again the relevance and goal of this study – drought risk, development for climate services. This is missing here.
• In addition, the conclusion is missing an outlook. What is still missing for the development of climate services? What is the applicability of this study going forward? What is the applicability of this study going forward?
• Last comment needs more elaboration. This section should not be a summary of results but really dive into the limitations and outlook etc.

More specific and editorial comments are given below:

Line 8-10: Is this conclusion really supported by the results at hand? I suggest to soften or to analyse a larger variety of individual events to support this statement.

Line 15: Improvement in observations? Unclear! Do you mean in the availability of observations or do you mean in the quality of available data such as reanalysis (as you call reanalysis observations)? Please rephrase. This statement can be short in the abstract but needs to be supported by more elaborations in the conclusion. (see comments above on conclusion)

Line 21: I suggest reformulating “soil surface” to terrestrial surface or similar.

Line 23: I suggest removing “at the surface” form the sentence

Line 23 – 26: Unclear sentence structure, I suggest reformulating.

Line 27 – 30: Sentence unclear, please reformulate.

Line 32 – 34: Remove “the” after drive: “drives 90% of the inter-annual variability”. Add more reference studies or soften the statement, e.g. ”the variability of soil moisture simulations has been found to drive (…)”

Line 38 – 41: Shorten the sentence and use same tenses throughout the text. Suggestion:  “In addition, Li et al. (2021) evaluated groundwater recharge estimations from different land surface models and found that the seasonal cycle of simulated groundwater storage (…).”

Line 47: Again, check for consistent usage of tenses. Which reanalysis products were compared?

Line 50 – 51: Please reformulate “regards (...)”. For which product/land surface model?

Line 55: Which land surface model? I suggest removing this part (Line 54 – 57) as it is not really relevant to this study and makes the transition to the next paragraph a bit confusing/abrupt. (Alternatively, more elaborations are needed here to make the transition to next paragraph more comprehensive.)

Line 63: I suggest reformulating to something like: “(…) soil moisture is one of the most impactful land parameters and is crucial for the forecast skill. “

Line 68: I suggest reformulating: “This can be attributed to reduced variability (..). “

Line 75: This statement is false! Boas et al. looked at soil moisture predicted with LSM that was forced with atmospheric fields of SEAS. Please rephrase.

Line 87: “(…) to wet and dry events. “

Line 90: Accordingly, the paper is structures as follows. I suggest to shorten this whole paragraph substantially. This is not needed in this detail for a scientific manuscript.

Line 105: “Second, the complex orography of this region (…). “

Figure 1: Not the best choice of colormap. Please consider using a colormap that conforms to color blind standards.

Line 130 onwards: Why not use the bias adjusted version of ERA5?

Line 137: Add reference citation for this statement.

Line 137 – 139: This belongs to methods section (is already mentioned there as well).

Line 152 – 153: “(…) with a mean water table depth below 10 m, (…).” Add reference citation.

Line 170 – 171, Equation is incomplete.

Line 183 – 184: Good statement but should be mentioned first in the introduction!

Line 200: I suggest reformulating this throughout the text: “1-month lead time”

Figure 2: Remove one “at “ and missing parenthesis in Figure caption: Columns show the same statistics for the forecast values at different forecast lead times (1, 3 and 6 months).

Figure 4: “(..) over the whole domain (…).” I suggest adding the soil layer depth for all layers either to figure or caption.

Line 233: I suggest rephrasing to avoid confusion: “(…) and only for shorter lead times. “

Line 242 – 244: Belongs to discussion/conclusion.

Figures 5 and 6: I suggest changing “Lead 1” etc. to Lead time – 1 months for example, to be consistent with the other plots and text. There were substantial differences in performance for the different layers, why not show wet periods for all layers?

Figure 7: Legend and axis names/labels are missing.

Line 256: (..) as shown in Figure 7b.

Line 257 – 259: Belongs to discussion.

Line 267: tab missing after “(Figure 8d)”.

Figure 8: Again, I recommend making adjustments to the figure caption, labels and titles, particularly ensuring consistency in lead time labels across all figures/captions and in the text.

Line 285: “nor”.

Line 287 – 288: I suggest: “The seasonal model analysed in this study is (..)”

Line 294 – 295: Replace “this paper” with this research/this study or similar.

Line 298: regions

Line 299: “(…), even when considering only the deepest layer; “

Line 301: regions

Line 302: coefficients

Line 301 – 302: Which regions?

Line 304: forecasts

Line 305: regions

Line 305 - 307: This statement needs to be softened as this study did not really predict future events, e.g. to: “This indicates (…)”.

Line 308 – 309: Which means? This section should not only include a summary of results but for each statement/bullet point a discussion/conclusion is needed.

Line 310 – 311: Same as above. Conclusion needed from this result.

Review comments for “Assessment of seasonal soil moisture forecast over central Mediterranean toward groundwater management” by Silvestri et al.

The authors have evaluated SEAS5 soil moisture forecast skills in the Mediterranean region. They found that the deepest soil layer (289 cm) has more skill than the upper soil layers. Hence, the authors conclude that deep soil layer forecast can be potentially valuable for groundwater management regionally.

The authors present a comprehensive review of the relevant literature. The manuscript is well written, and the figure quality is generally good. A novel contribution of this study is the skill in the deepest soil layer forecast. For these reasons, I liked the manuscript, which should eventually be publishable. I recommend the following revisions for the manuscript.

• Please show the sensitivity of your findings for the reanalysis data selected, which is highly dependent on the selected model (Kumar et al. 2019). Authors may consider alternative data sources, e.g., GLEAMv3 (Martens et al. 2017) and MERRA2 (Gelaro et al. 2017).

Related to the above point, I am somewhat supervised in seeing a lower skill in the upper layer soil moisture anomalies even at shorter lead times, e.g., 1 and 3 months (Figure 2). The root zone (0-1m) has a memory time scale ranging from 2-4 months; I was expecting a higher skill at the shorter lead times.

Additionally, it is unclear if the Authors used ERA5-Land soil moisture data for observations. There are clear differences between ERA5 and ERA5-Land soil moisture data, especially for deeper soil layers (Muñoz-Sabater et al. 2021).

• Process level understanding – please discuss biophysical reasons behind more skillful deeper layer soil moisture prediction in SEAS5. The authors may consider showing the memory time scale in each soil layer for the reanalysis of data and comparing the memory-based predictions with SEAS5 predictions.

• Figure 7 needs a thorough revision – the legend text is missing. X-axis labels are missing. Also, I would suggest two groundwater well data separate in (a) and (b), and they can be compared with the corresponding reanalysis data.

Detailed comments:

Title: ‘toward’ -> ‘for’

Line 5: ERA5 reanalysis -> Is this ERA5 or ERA5-Land?

Line 6-7: ‘good performance in the … deepest layer’ -> why?

Line 165 to 170: SSMA -> Eq1.  -> Add parentheses in the numerator.

A related comment is that forecast biases (drifts) are a function of the lead time and forecast initialization months (Kumar et al. 2014); it is unclear how you have incorporated these effects in the anomaly calculation. In particular, if you look at Fig. 9 (c), the forecast anomaly does not match the corresponding reanalysis data, even at the start of the forecast! Is it the effect of forecast drifts? If so, this can be easily removed using lead month and forecast initialization month-dependent climatology (Kumar et al. 2014).

Line 204: ‘this can be reconducted…’ -> what does ‘reconducted’ refer to. Please consider simplifying this sentence.

Line 205: ‘temporal oscillation’ -> ‘temporal variability’

Figure 1 and other figures, too: please consider using a color-blind-friendly color scheme. For example, I can not clearly distinguish between red and green colors. Additionally, in Figure 1, RMSE < 0.25 looks similar to the color ranging between 0.75 and 1.25.

Figure 2b,c, and f-> why there are stipplings on the Gary color areas; I am assuming they are statistically insignificant correlations.

Figure 2l,m,n -> why darker red areas ( ACC > 0.7) are not stippled, but the yellow area (0.2<ACC<0.4) are stippled.

Thank you.

References:

• Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2). J Climate, 30, 5419-5454.
• Kumar, S., P. A. Dirmeyer, and J. Kinter, 2014: Usefulness of ensemble forecasts from NCEP Climate Forecast System in sub‐seasonal to intra‐annual forecasting. Geophysical Research Letters, 41, 3586-3593.
• Kumar, S., M. Newman, Y. Wang, and B. Livneh, 2019: Potential reemergence of seasonal soil moisture anomalies in North America. J Climate, 32, 2707-2734.
• Martens, B., and Coauthors, 2017: GLEAM v3: satellite-based land evaporation and root-zone soil moisture. Geosci Model Dev, 10, 1903-1925.
• Muñoz-Sabater, J., and Coauthors, 2021: ERA5-Land: A state-of-the-art global reanalysis dataset for land applications. Earth System Science Data, 13, 4349-4383.

The main purpose of this study is to determine the predictive power of the SEAS5 system for seasonal soil moisture. The focus was the deepest soil layer at 289 cm in the central Mediterranean region. The accuracy of the SEAS5 (re)forecasts was compared against ERA5 reanalysis datasets, assuming that the ERA5 reflects realistic soil moisture conditions. The specific research question of the study was to predict recharge (stated as a flow toward groundwater – line 88,89) during dry and wet periods.

Improving our forecasting abilities for the water cycle components is a very important subject. The authors tackle this critical problem so that the subject is relevant and timely. The paper is well written. However, the main drawback of this manuscript is that the results were not adequately explained and discussed, leaving many unanswered questions. Therefore, the paper needs to be thoroughly revised, along with the inclusion of additional sections.

My main comments are listed below:

• One of this study's findings is that the forecasted and simulated soil moisture values within the lowest layers were found to be higher than those in the other layers. This is an interesting finding, but the potential reason remains unexplained. Is it because the soil moisture variations in the deeper layers are significantly less than in the surface layers? I would be interested in seeing the monthly soil moisture variations in these deep soil layers.
• Extensive agricultural activities exist in all the examined regions (Veneto, Umbria, and Naples). However, irrigation was not mentioned in the manuscript. Considering that irrigation may significantly impact both soil moisture and groundwater levels, providing an explanation of the potential implications of irrigation for soil moisture and groundwater level changes might be useful.
• Moreover, in section 2.4, it was stated that groundwater observations are used as a direct proxy to differentiate dry and wet events. However, the dry and wet periods may be observed in water table levels within different time frames. For example, dry periods may have more immediate consequences as the water table declines due to direct water extraction from the aquifers for irrigation purposes. On the other hand, the water table may have a more muted response to wet periods due to the slow vertical movement of soil moisture. While water table observations include these signals, neither ERA5 nor SEAS5 account for irrigation and only indirectly account for them as a result of data assimilation. Further clarification is needed on such connections and the implications of irrigation on the findings of the study.
• Although it is important that ERA5 and SEAS5 are independent estimates (i.e., using different initial conditions, data assimilation methods, etc, as stated in lines 137-139), using different soil parameters might lead to very different soil moisture results even though all other forcings are comparable. Please compare the soil hydraulic parameter distributions of both model approaches and explain the potential implications of any existing differences.
• Moreover, in section 3 (line 165), it was stated that both datasets are interpolated over a common resolution (0.25 degrees). However, the way in which this resolution change was handled is missing. Were the soil parameters accounted for during the interpolation? The same moisture amount may lead to different volumetric water contents in different soil textures.
• There is a disconnection between the main objective stated at the beginning of the manuscript and the findings of this study. The main research question is, “Can seasonal soil moisture forecasts be used to predict the flow toward groundwater?” However, the study did not attempt to predict groundwater recharge; rather, it sought to find the relationship between soil moisture trends and groundwater level changes in dry and wet periods. Please reword the main objective of the paper.
• Finally, since both the title and the main objective mention managing water resources, I was expecting some discussion about how this study's findings can be utilized for water management purposes, but such a discussion is missing.

Minor comments:

• Please include an explanation of dotted areas in the Figure 3 caption.
• Equation 1 is incorrect. Please add a parenthesis to the numerator to fix it.
• Figure 7 is missing legend and axis information.