Evaluation of SMAP Level 4 Versions 7 and 8 Soil Moisture Data in Rain-Fed Argentine Pampas Crops

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript evaluates the performance of the SMAP Level 4 soil moisture product (SPL4SMGP) versions 7 and 8 using in situ observations from two sites in the Argentine Pampas, focusing on Typic Argiudolls soils under rainfed agricultural conditions. The study provides a daily assessment of the 3-hourly product and compares product performance under dry versus normal-to-wet conditions.

The analysis offers useful insights into product behaviour across seasonal conditions; however, its representativeness is constrained by the limited number of sites, which are also located in close proximity. Additionally, some of the interpretations, including discussions on vertical coupling, remain largely speculative and would benefit from stronger supporting evidence.

I suggest major revisions, since several aspects of the scientific framing, methodology, and interpretation could be strengthened.

First, the spatial resolution of the SMAP product should be clearly stated from the beginning, throughout the manuscript (e.g., in the abstract and discussion), as it is essential for interpreting the comparison with point-scale in situ measurements. Related to this, the comparison between SMAP-derived root-zone soil moisture (0–100 cm) and in situ measurements (5–50 cm) is only clarified later in the manuscript. This mismatch in measurement depth is important.

Second, the manuscript would benefit from a more concise presentation. The Results and Discussion sections are currently quite wordy and contain substantial speculation that is not sufficiently supported by the analysis. In particular, interpretations regarding processes such as vertical coupling are not adequately substantiated with evidence.

Third, the study lacks sufficient contextualization within the existing literature. A comparison of the obtained results with previous validation studies of surface soil moisture (SSM) and root-zone soil moisture (RZSM) products in other regions and soil types would help to better position the findings and assess their broader relevance.

Fourth, the choice to focus on daily aggregated values of a 3-hourly product is not fully justified. It would be helpful to clarify why the daily assessment is preferred, or alternatively, to include both daily and 3-hourly evaluations to assess potential differences in performance.

Finally, the presence and treatment of bias requires further attention. It is unclear whether systematic biases between SMAP and in situ observations are quantified and, if present, whether any correction can be considered. Addressing this would improve the robustness of the evaluation.

Overall, the study has potential, but the analysis and interpretation would benefit from a more rigorous, concise, and better-supported approach. I list more specific comments here:

Figure 2: This figure can be improved by softening the gridlines. All the black lines make it busy and chaotic.

Line 175-176: “... to sense the relative dielectric permittivity, volumetric water content, ...” -- The statement suggests that volumetric water content is directly sensed, apart from dielectric permittivity, whereas in TDR-based systems it is derived from the measured dielectric permittivity. This could be clarified.

Line 184, 187: NASA Catchment land surface model = CLSM?

Table 1 is unclear and chaotic. Please improve readability. I propose to add horizontal lines or coloured rows.

Line 241 etc: To asses vertical coupling, I think you should look into temporal dynamics and the coupling over time.

Line 256: “lower variability of the deeper layer”? Figure 5 shows the deep variability is larger than the surface variability, at least for dry conditions.

Line 221-259: Very long part only on observed soil moisture distributions. Could be shortened. What is important here?

Subtitles would improve readability of the results and discussion section.

Figure 6:

• Improve the caption by indicating that SSM and RZSM sensor measurements and SMAP estimates are compared here. Add letters to the subplots for referral in the text.
• Lines can be misleading since you have missing data in your time series. As such, some of the straight lines are not real data. I suggest to use point markers instead.

Line 284 + 297 etc: SMAP RZSM is estimating 0-100 cm soil moisture, so it makes sense that this soil moisture is more buffered and less dynamic compared to the 5-50 cm depth measurement. Your sensor is not measuring Bt2 – C horizons.

Table 2 & 3 can be combined for the sake of space and readability.

Good luck!

Author Response

Dear Reviewers,

We would like to sincerely thank you for your valuable comments and suggestions on our manuscript. We believe that your feedback has significantly contributed to improving the quality and clarity of the work.

In response, we have carefully revised the manuscript, improving the language throughout as well as enhancing the figures and tables. All changes have been highlighted in gray within the revised version of the manuscript for ease of reference.

Below, we provide detailed, point-by-point responses to each of your comments.

Thank you again for your time and consideration.

Sincerely,

The Authors

Reviewer 1

The manuscript evaluates the performance of the SMAP Level 4 soil moisture product (SPL4SMGP) versions 7 and 8 using in situ observations from two sites in the Argentine Pampas, focusing on Typic Argiudolls soils under rainfed agricultural conditions. The study provides a daily assessment of the 3-hourly product and compares product performance under dry versus normal-to-wet conditions.

Q: The analysis offers useful insights into product behaviour across seasonal conditions; however, its representativeness is constrained by the limited number of sites, which are also located in close proximity. Additionally, some of the interpretations, including discussions on vertical coupling, remain largely speculative and would benefit from stronger supporting evidence.

A: We acknowledge that the analysis is based on a limited number of sites, which are located relatively close to each other (~23 km; lines 101 and 141). However, this limitation primarily reflects the scarce availability of in situ soil moisture data in the region. Although a regional monitoring network exists (Soil Moisture Telemetry Network of the SAOCOM Mission), it does not provide measurements over Typic Argiudolls at depths comparable to those analyzed in this study. In Argentina, available datasets are generally short-term, restricted to surface soil moisture, or correspond to different soil types, which hinders more comprehensive evaluations.

Despite the localized nature of the analysis, the selected sites are representative of the predominant conditions of the southeastern Pampas region. This area is characterized by relatively flat topography, homogeneous land use (mainly rainfed extensive crops such as soybean, wheat, and barley), and similar soil properties (Figs. 1 and 2), which reduces spatial variability at the pixel scale. Moreover, this homogeneity is supported by the land cover classification used as input in the product (MODIS MCD12C1), which classifies the area as “cropland.” Please see sections 2.1 and 2.2.

In addition, previous studies have shown that point-scale in situ measurements at sites such as those analyzed here can be effectively used to evaluate coarse-resolution satellite soil moisture products (e.g., SMAP or SMOS), despite spatial scale mismatches. This approach provides valuable insights into product performance in regions lacking dense observation networks [18,27,28].

Finally, the temporal series includes contrasting hydroclimatic conditions (dry and normal–wet campaigns), strengthening the evaluation by covering a relevant range of soil moisture dynamics for the soil type considered.

On the other hand, Section 3 has been modified to avoid speculative discussions.

I suggest major revisions, since several aspects of the scientific framing, methodology, and interpretation could be strengthened.

Q: First, the spatial resolution of the SMAP product should be clearly stated from the beginning, throughout the manuscript (e.g., in the abstract and discussion), as it is essential for interpreting the comparison with point-scale in situ measurements. Related to this, the comparison between SMAP-derived root-zone soil moisture (0–100 cm) and in situ measurements (5–50 cm) is only clarified later in the manuscript. This mismatch in measurement depth is important.

A: Information on spatial resolution was added to the abstract (line 21). In Section 2 (Materials and Methods), subsection 2.2, lines 138–142, the following clarification was included: Although measurements correspond to different sites and thus different pixels, it should be noted that this product is the result of interpolating the brightness temperature measurements from the SMAP footprint (~40 km) on a 9 km grid [4-6]. 

We appreciate this interesting comment and we agree that the mismatch in monitoring depths needs attention. We support the carried out comparison considering the lack of SM data for the deepest horizons in the study area, profile homogeneity at depths greater than 50 cm, and following the method of different previous studies (section 2.3). Some studies validating SMAP products (e.g. [27], Albergel et al., 2012) reported that SM up to 50 cm can be considered representative of the 0−100 cm depth soil moisture. Additionally, for this study, only a single SoilVUE™10 sensor was available. However, we acknowledge that there may be some fluctuations that we are not taking into account, and the reliability of the results obtained could be partially affected. This point has been added to the text.

Lines 278-281: Results suggest that the model smooths short-term fluctuations and attenuates rapid drying signals driven by atmospheric forcing at the surface. However, part of these discrepancies can be associated with the different depths considered [27].

Conclusions: Although the analysis was carried out taking into account previous studies monitoring the soil profile at 0-50 cm depth, some soil moisture dynamics may be under-monitored. It should be noted that the lack of data in the study area did not allow for an analysis of deeper horizons. Those future studies should complement measurements at 50-100 cm depth for a more comprehensive analysis of the product

Albergel, C., de Rosnay, P., Balsamo, G., Isaksen, L., Muñoz-Sabater, J., 2012. Soil moisture analyses at ECMWF: evaluation using global ground-based in situ observations. J. Hydrometeorol. 13, 1442–1460. https://doi.org/10.1175/JHM-D-11-0107.1.

Q: Second, the manuscript would benefit from a more concise presentation. The Results and Discussion sections are currently quite wordy and contain substantial speculation that is not sufficiently supported by the analysis. In particular, interpretations regarding processes such as vertical coupling are not adequately substantiated with evidence.

A: We appreciate this constructive comment. Section 3 has been revised to improve conciseness and to avoid speculative discussions. Interpretations that were not sufficiently supported by the analysis have been removed or carefully rephrased to better reflect the evidence provided.

Q: Third, the study lacks sufficient contextualization within the existing literature. A comparison of the obtained results with previous validation studies of surface soil moisture (SSM) and root-zone soil moisture (RZSM) products in other regions and soil types would help to better position the findings and assess their broader relevance.

A: We thank the reviewer for this valuable suggestion. The bibliography has been expanded, and the results are now discussed in comparison with findings from studies conducted in other regions. These changes have strengthened the contextualization and interpretation of our results. Please see Lines 64-83 and Subsection 3.2, lines 305-317. 

Q: Fourth, the choice to focus on daily aggregated values of a 3-hourly product is not fully justified. It would be helpful to clarify why the daily assessment is preferred, or alternatively, to include both daily and 3-hourly evaluations to assess potential differences in performance.

A: Our evaluations were based on metrics computed at a daily scale, as intraday variability was not significant and the statistical results were similar to those obtained using 3-hourly data. The choice of a daily timescale is also consistent with relevant changes in the soil-plant system for agricultural applications in the study area

While sub-daily analyses may be relevant for more detailed hydrological studies (e.g., to assess runoff dynamics or intraday variability), such approaches are beyond the scope of this work. Therefore, the use of daily data is considered appropriate in relation to the soil–plant interactions meaningful in a context of productive large plains. The justification has been improved in the revised manuscript: Subsection 2.3, lines 196–201.

Q: Finally, the presence and treatment of bias requires further attention. It is unclear whether systematic biases between SMAP and in situ observations are quantified and, if present, whether any correction can be considered. Addressing this would improve the robustness of the evaluation.

A: The systematic and random errors were evaluated by comparing the field and SPL4SMGP SM data. Section 2 and the heading of Table 2 have been modified to clarify this point. Corrections of errors were not carried out because the aim was a direct comparison with field data to evaluate the SMAP product performance. Table 2 shows that, in general, the bias (systematic error associated with overestimation) is similar to or higher than random errors (ubRMSD), and they are comparable with the SMAP mission objective (≈0.04 m³/m³). Thus, correcting the data would not provide significant improvements. During dry periods, future studies should confirm the observed higher bias.

Overall, the study has potential, but the analysis and interpretation would benefit from a more rigorous, concise, and better-supported approach. I list more specific comments here:

Q: Figure 2: This figure can be improved by softening the gridlines. All the black lines make it busy and chaotic.

A: Thank for the observation. Figure 2 (now Figure 1) has been revised in accordance with the reviewer’s suggestion.

Q: Line 175-176: “... to sense the relative dielectric permittivity, volumetric water content, ...” -- The statement suggests that volumetric water content is directly sensed, apart from dielectric permittivity, whereas in TDR-based systems it is derived from the measured dielectric permittivity. This could be clarified.

A: We thank the reviewer for noting this. The text has been revised accordingly (lines 147–150). 

Q: Line 184, 187: NASA Catchment land surface model = CLSM?

A: We appreciate the suggestion and have addressed it accordingly.

Q: Table 1 is unclear and chaotic. Please improve readability. I propose to add horizontal lines or coloured rows.

A: We agree with the reviewer’s observation, and Table 1 has been revised accordingly. 

Q: Line 241 etc: To asses vertical coupling, I think you should look into temporal dynamics and the coupling over time.

A: We thank the reviewer for the comment. We agree with the observation and have completely revised Section 2.3 and the discussion about vertical coupling analysis accordingly. 

Q: Line 256: “lower variability of the deeper layer”? Figure 5 shows the deep variability is larger than the surface variability, at least for dry conditions.

A: The sentence was removed as it was considered confusing. 

Q: Line 221-259: Very long part only on observed soil moisture distributions. Could be shortened. What is important here?

A: We thank the reviewer for the comment. The focus of this section and the analysis have been reorganized accordingly. 

Q: Subtitles would improve readability of the results and discussion section.

A: We thank the reviewer for the comment. We agree with the observation, and subheadings have been added to Section 3 accordingly. 

Figure 6:

• Improve the caption by indicating that SSM and RZSM sensor measurements and SMAP estimates are compared here. Add letters to the subplots for referral in the text.
• Lines can be misleading since you have missing data in your time series. As such, some of the straight lines are not real data. I suggest to use point markers instead.

A: We appreciate the suggestion. The figure caption has been revised, and the subplots have been labelled with letters. Markers were also incorporated as suggested. Additionally, following Reviewer 2’s recommendation, the time series for barley and wheat have been included.

Q: Line 284 + 297 etc: SMAP RZSM is estimating 0-100 cm soil moisture, so it makes sense that this soil moisture is more buffered and less dynamic compared to the 5-50 cm depth measurement. Your sensor is not measuring Bt2 – C horizons.

A: Please see the response to question 2.

Q: Table 2 & 3 can be combined for the sake of space and readability.

A: We appreciate the suggestion. Tables 2 and 3 have been merged accordingly.

Reviewer 2 Report

Comments and Suggestions for Authors

Soil moisture (SM) is a fundamental variable in agricultural and natural ecosystems. SPL4SMGP provides large scale and continuous SM data, facilitating the monitoring of its variability. Validation and calibration of satellite-derived soil moisture products is important for understanding the performance and further improving the algorithm.

This manuscript evaluates the performance of SPL4SMGP SSM and RZSM using field observations from the Argentine Pampas, with normal-wet and dry conditions. Results showed both SSM and RZSM achieved ubRMSD values close to the SMAP accuracy target, while SSM correlated moderately with observations but RZSM displayed weak sensitivity to soil profile variability, and Version 8 showed similar performance to version 7, with a tendency towards overestimation.

The manuscript presents the results and analysis clearly and understandably. However, the methods and materials are lack of novelty, with point-pixel comparison and traditional evaluation metrics. Here are some suggestions to further enhance the manuscript.

Main concerns:

1. The field data, “collected by a sparse network”, only has 2 sites (not really a network). We don’t get the information how faraway these 2 sites are, and whether they are located in one pixel. Regarding the representative of the field data, I recommend to use remote sensing SM index, or carry out a few observations with hand-hold instruments, to show the SM spatial distribution in the study pixel.
2. As the authors mentioned, the precipitation and soil properties are two main factors which affect the performance of the SPL4SMGP products. But the analysis is qualitative instead of quantitative, and the latter is more useful for improvements of the algorithm. The fig. 6 shows there are at least 3 false precipitations during the dry period of 2020-2021, there should be more in other dry periods. It would be interesting to get the information how the bias of the forcing data affects the SM estimation.
3. Fig. 6 shows the temporal series of SSM and RZSM, during two dry and normal-wet soybean growing seasons, what about that of Barley and wheat growing seasons, especially the 2022 and 2023?

Specific comments:

1. Line 65-67, please add more relevant references, instead of only one from the authors.
2. Line 101, “the same crop8”, as if there is something missing.
3. Line 175, vertically – horizontally?
4. Table 1, is it better for listing the difference between version 7 and 8?
5. Line 256, “like evapotranspiration processes21”, [21]?
6. Fig. 5 and fig. 6, there are some SM observations lower as zero. Regarding the high organic matter and clay loam texture, I don’t think the TDR could get SM as low as zero at 5cm depth. 

Author Response

Dear Reviewers,

We would like to sincerely thank you for your valuable comments and suggestions on our manuscript. We believe that your feedback has significantly contributed to improving the quality and clarity of the work.

In response, we have carefully revised the manuscript, improving the language throughout as well as enhancing the figures and tables. All changes have been highlighted in gray within the revised version of the manuscript for ease of reference.

Below, we provide detailed, point-by-point responses to each of your comments.

Thank you again for your time and consideration.

Sincerely,

The Authors

Reviewer 2

Soil moisture (SM) is a fundamental variable in agricultural and natural ecosystems. SPL4SMGP provides large scale and continuous SM data, facilitating the monitoring of its variability. Validation and calibration of satellite-derived soil moisture products is important for understanding the performance and further improving the algorithm.

This manuscript evaluates the performance of SPL4SMGP SSM and RZSM using field observations from the Argentine Pampas, with normal-wet and dry conditions. Results showed both SSM and RZSM achieved ubRMSD values close to the SMAP accuracy target, while SSM correlated moderately with observations but RZSM displayed weak sensitivity to soil profile variability, and Version 8 showed similar performance to version 7, with a tendency towards overestimation.

The manuscript presents the results and analysis clearly and understandably. However, the methods and materials are lack of novelty, with point-pixel comparison and traditional evaluation metrics. Here are some suggestions to further enhance the manuscript.

Main concerns:

Q: The field data, “collected by a sparse network”, only has 2 sites (not really a network). We don’t get the information how faraway these 2 sites are, and whether they are located in one pixel. Regarding the representative of the field data, I recommend to use remote sensing SM index, or carry out a few observations with hand-hold instruments, to show the SM spatial distribution in the study pixel.

A: We appreciate this insightful comment. The information about sites and data has been clarified (Subsection 2.2).

About the representativeness of field data, we provided justification for that aspect in Subsection 2.2. In this sense, the SMAP SM product incorporates land cover information that classifies the study area as homogeneous cropland, supporting the representativeness of the selected sites. The two monitoring locations were considered to capture crop and soil conditions representative of the landscape scale based on the comparisons between field and satellite data carried out in previous studies [18, 22, Niclòs et al., 2013, Holzman, et al., 2017]. These previous studies conducted in the study area have evaluated the representativeness of site-specific SM through inter-site comparisons and validation against coarse-resolution SM products. Although the measurements correspond to two different sites and different pixels, it is important to note that the SMAP product is derived by interpolating brightness temperature observations from a footprint of approximately 40 km onto a 9 km grid [4–6].

Given the proximity between sites (~23 km) and their similar soil, climate, and crop conditions, all field data were analyzed jointly. In addition, previous studies in the region (e.g., Mancino and Rivas, 2021) reported strong agreement among SM of nearby stations (R² > 0.91), suggesting that under relatively homogeneous edaphic conditions, spatial variability in SM is limited. Therefore, the point-scale measurements used in this study can be considered reasonably representative of the average signal of the SMAP footprint. (Subsection 2.2, lines 136-145).

Niclòs, R., Rivas, R., García-Santos, V., Doña, C., Valor, E., Holzman, M., Bayala, M., Carmona, F., Ocampo, D., Thibeault, M., Soldano, A, 2013. SMOS soil moisture product validation in croplands. Poceedings of ESA Living Planet Symposium, SP-722, 9 a 13 de septiembre de 2013, Edimburgo-Reino Unido. ISSN 1609-042X. DOI: 10.13140/2.1.5010.1442 http://www.livingplanet2013.org/abstracts/837074.htm

Holzman, M.E., Rivas, R., Carmona, F., Niclòs, R., 2017. A method for soil moisture probes calibration and validation of satellite estimates. MethodsX, 4: 243-249. http://dx.doi.org/10.1016/j.mex.2017.07.004 (ISSN 2215-0161)

Q: As the authors mentioned, the precipitation and soil properties are two main factors which affect the performance of the SPL4SMGP products. But the analysis is qualitative instead of quantitative, and the latter is more useful for improvements of the algorithm. The fig. 6 shows there are at least 3 false precipitations during the dry period of 2020-2021, there should be more in other dry periods. It would be interesting to get the information how the bias of the forcing data affects the SM estimation.

A: We agree that these factors are important to the product's performance. As mentioned in the first question of the first reviewer, a detailed and more quantitative analysis covering different soil types and textures in the study area is not possible, given the lack of subsurface SM data. However, the aim was to evaluate the performance of the SMAP product in the scarcely studied Typic Argiudolls of the Argentine Pampas. In this context, the discussion was more focused on the forcing factors and some aspects that arise from the analysis of input data (lines 335-351). Results suggest that improvements in the model parameterization could yield better results. Also, in the Conclusions, we have added text about the limitations of the study and future lines of research.

About the comment on precipitation in Fig. 6, we could not fully understand the question. The shown precipitation includes field measurements. Some atypical cases of increases in SM with rainfall not observed in the field may occur due to the spatial heterogeneity of some rainfall events. However, it can be observed that rainfall pulses are associated with increases in SM. Some very small precipitation pulses (in general, lower than 5 mm) have no effect on SSM due to processes such as interception by vegetation and evaporation.

1. 6 shows the temporal series of SSM and RZSM, during two dry and normal-wet soybean growing seasons, what about that of Barley and wheat growing seasons, especially the 2022 and 2023?

A: The suggestion was taken into account, and the analysis of the barley and wheat campaigns for 2022 and 2023 was carried out. Lines: 271-289

Specific comments:

1. Line 65-67, please add more relevant references, instead of only one from the authors.

A: The suggestion was taken into account, and a more comprehensive review of the literature was conducted. Lines: 64-83.

2. Line 101, “the same crop8”, as if there is something missing.

A: The subsection has been revised, and the sentence has been removed.

3. Line 175, vertically – horizontally?

A: The SoilVUE™10 is installed vertically. It includes sensors at different depths that perform horizontal measurements (please see Fig. 2). Lines 147–150.

4. Table 1, is it better for listing the difference between version 7 and 8?

A: Table 1 summarizes the main differences and similarities between the two versions, to help the reader identify the changes introduced. In addition, following the suggestion of Reviewer 1, the table has been revised to improve its readability and clarity. 

5. Line 256, “like evapotranspiration processes21”, [21]?

A: We appreciate the comment. The reference has been modified. Lines 255.

6. 5 and fig. 6, there are some SM observations lower as zero. Regarding the high organic matter and clay loam texture, I don’t think the TDR could get SM as low as zero at 5cm depth. 

A: We thank the reviewer for the observation. We agree that negative soil moisture (SM) values are not physically plausible in the considered soils of the study area. These values are associated with occasional periods of TDR sensor instability and/or calibration mismatches, possibly linked to temperature effects, variations in soil electrical conductivity, or imperfect sensor–soil contact in the upper layers.

Although quality control procedures were applied to the in situ data (including range checks and visual inspection of the time series), a small number of unrealistic values remained in the dataset. In response to this comment, an additional filtering step was implemented, removing all near-zero values (< 0.03 m³/m³) before the analysis (Subsection 2.3, paragraph 2).

This adjustment does not affect the overall results, as these anomalous values represent a very small fraction of the dataset and do not show a consistent temporal pattern. The manuscript has been revised to explicitly describe this procedure (Subsection 2.3, paragraph 2), and Figures 5, 6 and Table 2 have been updated accordingly.

Additionally, regarding the graphical representation (now Figure 5), the “tails” extending toward zero or slightly negative values in the violin plots are an artifact of the kernel density estimation (KDE) method used to construct these plots, rather than actual data values. This smoothing approach can artificially extend the distribution beyond the observed range, particularly near the boundaries.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript has been improved significantly based on the comments and suggestions.

Author Response

Dear Reviewer,

Thank you very much for your positive evaluation of the revised manuscript. We sincerely appreciate your time and effort throughout the review process.

Sincerely,

The Authors

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks to the efforts of the authors, I believe most of my concerns have been addressed.

Regarding the comment on precipitation in Fig. 6, what I mean is that there are SM pulses in the SMAP products, but they were not observed by ground instruments, which should be caused by false precipitation in the forcing data. It would be interesting to assess how the bias of the forcing data affects the SM estimation.

Regarding Fig. 7 and Fig. 8, please add a description of the data distributions diagram in the figure caption.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf