The authors have examined top soil moisture and the myriad of environmental and climatological variables controlling it in a well-studied catchment area in northern Sweden. The study design includes a good amount of top soil moisture measurements and studies the impact of different variables, as well as their resolution/calculation methods, on the spatial and temporal variation of soil moisture. Overall, the manuscript is well executed and the study design interesting. However, there are a few smaller and bigger aspects of the paper that would benefit from some additional work.

General comments:

Abstract:

It was somewhat difficult to get the main points of the article from the abstract. The beginning is very broad and so are the descriptions of what precisely was done. I understand that there were a very large number of variables involved so summarizing all relevant aspects is not feasible but perhaps a bit more precision would help. I also find the reasoning for a boreal forest site a bit lacking, surely there are other reasons to look into soil moisture in these areas other than lack of data?

Introduction:

I’m not entirely convinced about the main goal of this paper and how it is presented. For one, this very much reads like an empirical modelling paper, but a lot of emphasis in the introduction is placed on understanding the mechanisms and processes driving soil moisture. These would, in my opinion, be better studied with more process-based methods such as mechanistic models or field research focusing on the processes themselves instead of the proxies describing them (such as topography-related indices describing flow patterns in a landscape).

Secondly, I understand the benefit of looking into a multitude of different variables at the same time. However, the good side of being more selective with your variables is that you then have to justify them properly and this is where I think the paper is currently lacking. There are some variables that seem to describe the same thing such as two variables for vegetation biomass and two datasets for soil properties without much justification while some aspects are ignored (such as the topographical variation in radiation). Some spatial variables are tested in multiple resolutions while others are not. This problem also reflects to the results and discussion. As the amount of variables is large and the reasonings behind them a bit unclear, it is challenging to cover and understand all the relevant findings. For example, if one of the goals is not to look at how different datasets of the same variables fare (such as SLU vs. SGU soil data and ERA-5 vs. field data), then why include multiple datasets? This is in my opinion one of the very interesting questions in this type of analysis, yet it is ignored.

Methods:

I again appreciate the multitude of variables but I do not think they are sufficiently covered in the method section. It is not enough cite previous papers without providing almost any explanation of what the variables are and how they have been defined. It makes it nearly impossible for the reader to estimate if your results are reasonable and expected when the reader can’t know what was measured without going through various papers, some of them in a foreign language (SLU). Perhaps I missed it, but I’d also like to know the original resolution of the raster datasets.

Would it be possible to provide at least a few maps of the main variables for example in the supplement so that the reader can get a better understanding of the catchment? For example topography, vegetation, land cover and soil type would already provide a lot of very useful information.

Discussion:

There is in general throughout the article very little discussion of how the site characteristics influence the results and how well these are applicable outside this study area. I’d also pay a bit more attention to why certain results are as they are and be clear in communicating them. For example, in L420, the longer-term effect of soil temperature is likely due to the fact that soil temperature at those depths (28-100 cm) also varies slowly compared to the top soil temperature. While this is rather obvious, it’s maybe good to point it out. Similarly, in L444-446, I would spell out more clearly how vegetation patterns impact soil moisture. This can be for example due to increased transpiration during peak growing season or the impact of shading. Daylengths and their temperatures are not very clear explanations.

Furthermore, there are clearly things that are not measured here, that would influence soil moisture variation, for example the spatial variation of meteorological variables and I do think acknowledging those in the discussion is important.

Specific comments:

You refer several times to your study period as vegetation period. I’m not familiar with the term so could you define it?

Introduction:

L49: “All potential controls” is a very ambitious term and I’m not entirely sure it is, or can be, achieved with black-box models (or with process-based ones either) considering the interplays of soil moisture with many of its predictive variables, the often massive heterogeneity of soil properties and the need for proxy variables such as topographical indices. While I appreciate the scope of this study, I would perhaps phrase this differently.

L69-71: In relation to the comment above, this is a much clearer version of the same sentence. However, I’m not sure both of these are needed in the same introduction.

L78: I understand that the cited papers don’t cover areas outside boreal forests and subarctic tundra, but surely this same thing is true in any cold climate with a seasonal snow cover?

L83: “However, recent research indicates that topography may have a different relationship with soil moisture under varying wetness conditions.” This is a rather vague sentence. Do you mean to say that the impact of topography differs depending on the wetness conditions?

Methods:

L116: Is the catchment area primarily managed boreal forest and if yes, how is it managed? I could imagine that managed boreal forests differ in their soil moisture controls compared to non-managed forests so this could at least be mentioned somewhere.

L132: I fully understand the separation of the variables into spatially and temporally varying ones. However, it would be good to somewhere, for example in the discussion, recognize that many of the temporal variables are indeed not spatially homogeneous. For example air temperature, particularly close to the ground, can vary considerably (and is often tightly connected to soil moisture), transpiration naturally depends on vegetation, radiation on the topography, etc.

L135: How was the subset selected?

L150: Nothing to correct here, just wanted to say well done for adequately explaining how you did the calibration!

Results:

L239: It might be worth noting that the sharp decline during precipitation events starts happening after in August. In July the responses are very small. I would also perhaps use the term “precipitation event” instead of “precipitation occurrence”.

L278: This is very nit-picky, but could you place the abbreviations of plan curvature and downslope index other way around so they’re consistent with the rest of the sentence?

L281: This is a good example of why explaining the variables in more detail and justifying the selection would be beneficial. Now it very much seems that you’re trying to explain soil moisture by examining soil moisture, while it probably is just interesting to see how well these two correspond with each other. The same goes for the pine variables in the next paragraph (L289 and L290).

L294: The end of the sentence is missing something.

Discussion:

L373: I’d be careful when using the word predict. In my understanding, this type of modelling is trying to explain the variation, since there aren’t predictions outside the measurements.

L395: I’m not entirely convinced that Kemppinen’s study site is all that comparable with the Krycklan catchment considering the difference in vegetation (treeless tundra vs. boreal forest) but it’s also very difficult to say since there are little maps providing information on the characteristics of your cathcment. As an interesting side note, having visited the valley, I’d suspect that the reason for TWI being more useful there is due to the shape of the valley which very strongly gathers the water flow to low-lying areas (and there are also deep organic layers at the bottom of the valley due to this, further enhancing the accumulation of soil moisture). This is probably a good example of exactly what you also show in the paper, that the characteristics of different watersheds are important.

L399: “...their spatial resolutions and thresholds. We argue that…”

L420: It might be useful to point out that such a deep soil temperature also fluctuates very slowly compared to for example top soil temperature. Furthermore, finer spatial resolution of air temperature might have yielded different results.

L442: “Regarding vegetation, we did not find a direct evidence...”

L443: I’m not sure that the article by Teuling et al is very useful here. First of all, it looked at evapotranspiration driven by soil moisture, not the other way around. This is somewhat semantics but I do think it’s good to remember which processes drive which (or if they are driving one another). Secondly, and this would be interesting to study further, the Teuling-article studied single points in various ecosystems whereas you’re concentrating on much more fine-scale variation of soil moisture. These might not behave in a similar way.

L447: Could you be a bit more precise here with the word “differently”?

Figures:

Fig. 1: Ignore this comment, if you think it’s not suitable, but would it be possible to get the main streams within the catchment area visible on the map?

Fig. 2: Overall an informative figure, but could perhaps the arrows on “no trend” sites be removed for clarity? Also, in the legend of 2c, the symbol of ERA5-Land is indistinguishable from the other lines, it might help making the lines in the legend somewhat thicker than the actual lines in the plot.

Fig. 3: Should the resolution be in meters or in square meters?

The authors investigated the spatiotemporal controls of soil moisture in boreal forest landscape in Sweden. They monitored surface soil moisture at 82 locations during a few months period. These soil moisture measurements were analyzed together with a vast array of environmental and hydrometeorological factors. Different spatial and temporal scales were considered. The study contributes to advancing models that represent spatiotemporal soil moisture variability.

Overall, the scope of the study is important and suitable for HESS. The methodology is clear and rather straightforward, and the number of considered predictors is impressive. The paper is generally well conceptualized and written. However, I have some suggestions to further improve the paper, especially the introduction and discussion.

General comments:

In my opinion, the most concrete contribution that the study can make is to identify and provide key spatial and temporal predictors for data-driven models used in soil moisture mapping. While this aim and contribution is relevant and briefly mentioned, it would benefit from being more explicitly emphasized and clarified throughout the manuscript, particularly in the abstract, introduction, and discussion. Strengthening these sections could better position the study within the broader context of soil moisture research and mapping.

It remains unclear whether the study contributes to understanding the processes driving soil moisture variability. Therefore, it is also doubtful whether the study offers a contribution to the development of physically based models. Such models already incorporate key spatial (e.g., vegetation, soil texture, topography) and temporal (e.g., meteorological forcing) factors, and the manuscript does not address how the findings could enhance these models. Perhaps the authors could emphasize the need for more accurate field and remote sensing data on the identified variables, which could indirectly benefit physically based model approaches that rely on such data.

Although this study may not directly advance our understanding of the processes driving soil moisture variability, the identification of key predictors, many of which are already incorporated in physically based models, presents an opportunity for the authors to discuss how these different methods (process-based and data-driven) could be integrated for soil moisture mapping (e.g. hybrid approaches).

The amount of predictors and scales is impressive, but there are still many limitations to this study (as to any study). It is not a major problem, but the limitations need more attention in the discussion. For instance, topographic and tree shading is neglected, the spatial scales go only up to 30 or 64 m, and temporal extend was only a few months, just to name a few.

Specific comments:

L33-34: Please add reference to “influence soil nitrogen availability”

L36-39: It would help to have example reference to each of these applications that require soil moisture state.

L77-80: This is a good point that topography can have a major role in the early summer. In this study, you basically skipped the period impacted by snowmelt. How do you think the results would differ if full snow-free season was included?

L83-84: I don’t quite understand this sentence. You already described how the impact of topography changes between wet and dry seasons. Perhaps you mean that this may change within the season as well. This could be reformulated. Also please add references to the recent research you’re referring to.

L135: How were the soil moisture locations decided? They mostly fall in the intensitve monitoring area. How much of them are on peat soils? How much in mineral? Why this subset of the measurements?

L137: I was surprised not to see any fully saturated measurements on peat soils (peat soil porosity is around 0.90). Were there no peatlands (or peatlands with measurements) where the water table was near the surface? Hence, I don’t think these measurements “capture the full spectrum of soil moisture levels across the Swedish landscape”.

L167.: While citing for details is understandable, it would be good to elaborate a bit more on how the topographic indices were defined. For example, it makes a difference which method is used (e.g. D8 vs. Dinf). Please describe also the site quality index.

Sect. 2.2.2: Please consider adding spatial predictor figure in supplement. It would give the reader a better idea how the landscape looks like. It is an impressive number of predictors, but one obvious one is missing: topographic shading.

Sect. 3.2. repeats many of the things already described in the methods (Sect. 2.3., which is great by they way). Perhaps the chapter in 3.2. can be integrated to 2.3.?

L349: What is a vegetation period? Analysis starts from July something, growing season starts earlier.

L370: “forest boreal landscape” -> “boreal forest landscape”

L394: Kemppinen et al. study site (tundra) is quite different to Krycklan.

L420: Consider replacing “performed poorly” with “correlated poorly with soil moisture” or something similar.

Code and data availability: Better practice would be to share the data in an openly available repository. Or has the data already been shared in the cited literature?