An Empirical Orthogonal Function-Based Approach for Spatially- and Temporally-Extensive Soil Moisture Data Combination

Round 1

Reviewer 1 Report

The topic of the manuscript is very interesting and fits the topics of Water.

The introduction section is very informative, and could form part of a review article. The introduction well addresses the problems that may be encountered when using soil moisture information and trying to correlate them with other variables.

The authors use empirical orthogonal function analysis to combine heterogeneous dataset formed by soil moisture data acquired manually and automatically. This analysis produces convincing results when applied to their datasets, which I believe are based on a large number of data that not all the sites can reach. Indeed, I don’t think that having 36 measurement days and 65 annual measurement locations is a so reduced dataset.

The question is then: how can such an approach be generalized to sites with less monitoring points or shorter duration of data acquisitions (not all the sites are CZ observatories). Could they sub-sample their dataset to determine the minimum number of data that is required to get a robust use of the EOF analysis? Even with a large dataset, they report some discrepancies between manually acquired and automatically acquired data and pointed out that data should be somewhat in agreement to be used (line 347). Consequently, is this method adapted to reduced datasets?

Apart from this specific aspect, the paper is well written and informative, the procedure used for interpreting the data is well described, and the potential outcomes of the proposed methods are also well addressed.

I recommend to accept the paper with moderate revisions, provided the potential use of the EOF approach with reduced datasets is discussed.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Manuscript Number: WATER-949569

Authors presented a methodology for assessing spatial and  temporal soil moisture variation by using the Empirical Orthogonal Function and integrating soil moisture manual measurements with auto-recorded datasets and topographic variables.

The topic is quite interesting and different datasets are used. Whilst the manuscript is reasonably well written, there are some issues that I feel the authors need to consider, particularly concerning methodology (detailed below) and consequently results. 

In the following, there are the main points on which I believe the authors need to supply more information or consider. For minor suggestions, please, see the attached pdf file. 

1. Abstract

I suggest the Authors to refine the abstract. In particular, the aims of the paper should be clearly state: to derive the dominant soil moisture patterns based on spatial-extensive datasets with temporal-extensive datasets and assess their relationships with terrain attributes and soil properties.

Apart from highlighting the proposed integrated approach, according to me, the abstract should be enriched with the main findings concerning soil moisture in the study area (e.g., soil moisture distribution in the study area is a function of both topography and soil depth; a weak fitting between manual measurements and auto-recorded soil moisture datasets was detected; etc...).

2. Material and methods: study area

A short overview of the geological background is required. Additional information concerning soil types and features of the soil five series (L.113-114) should be provided.

3. Data collection

What were the dates/season for data collection in the field? According to me, it should be explicitly stated. The season might make a difference to interpretation of the results.

4. Material and methods: Correlation coefficients

Pearson's correlation coefficient (L.192-194) is valid only if variables follow a normal distribution. Environmental variables usually fail such assumption. If soil properties are not transformed into Gaussian-shaped variables, a non-parametric coefficient requires to be used. The Spearman coefficient (ρ) gives insight about the strength of pairwise correlation whether the relationship between the variables is linear or not (cfr. Spearman, 1906: Footrule for measuring correlation. Br J Psychol 2:89–108).

5. Material and methods: Spatial interpolation

The procedure for spatial interpolation is completely missing in the methodology section although caption of fig. 2 clearly states “Krige maps…” and fig. 3 shows some variograms. It is very important for the reader to know how data have been processed.

Kriging refers to a family of different interpolation techniques (e.g. ordinary kriging, regression kriging, cokriging, kriging with external drift, etc..). Which one have you used?

Given that variogram modeling is sensitive to strong departures from normality, have data been transformed into a Gaussian-shaped variable before applying kriging? Which kind of transformation has been applied? In addition, it is recommended to provide the kriging variance or its square root (the kriging error) for each estimated variable since it highlights the reliability of the estimates.

6. Results and discussion

Results and discussion require to be modified once the previous comments n. 4, 5 are addressed.

7. Results: organic carbon content

In Table 1, organic carbon content appears (Table 1, L.245). Have you measured SOC? Such information is missing in the methodology section. Please add the details of the procedure followed for its assessment in the appropriate section.

8. Figures

From the caption of fig. 2 it seems that morphometric attributes (fig. 2a, b, d, e, f) derive from a kriging interpolation but they are easily derived from a DEM. Please modify the caption.

Legend of figs. 2 and 3 are too small to be read.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Dear Authors,

Thank you for answering my comments and for accepting my suggestions. The manuscript has been significantly improved however a couple of issue still remain.

The request to provide the kriging variance or the kriging error for each estimated variable (comment n.5 in my previous review report) has not been addressed. Hence, I recommend the Authors to enrich the manuscript with such information in order to evaluate the reliability of the estimates.

The legend of fig. 2 has not been enlarged and it is not enough clear to be read (comment n.8 in my previous review report). Please, modify the figure so that the reader can appreciate the values in the legend.

Just few refinements are annotated on the pdf file.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.