the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Jan 2025
Influence of Snow Spatial Variability on Cosmic Ray Neutron SWE
Abstract. Monitoring prairie snow has been difficult due to its extreme spatial variability from windy conditions, gentle topography, and low tree cover. Previous work has shown that a noninvasive (or aboveground) Cosmic Ray Neutron Sensor (CRNS) placed at the Central Agricultural Research Center (CARC; 47.07° N, 109.95° W), an agricultural research site within a semi-arid prairie environment managed by Montana State University, was sensitive to both the low snow amounts and spatial variability of prairie snow. In this study, we build upon previous work to understand how different snow distributions would have influenced CRNS measurements at the CARC. Specifically, we compared the changes in neutron counts and snow water equivalent (SWE) after relocating our CRNS probe at the CARC using the Ultra Rapid Neutron-Only Simulation (URANOS) and comparing them to uniform snow distributions. For shallow, heterogeneous snowpacks like the ones observed at the CARC, the magnitude and distance of the snow drifts from the CRNS has the greatest effect on neutron counts. Therefore, the best place to site a CRNS is within areas of low snow accumulation that are nearby areas of high snow accumulation to obtain a reasonable spatial estimate. Despite this, a naive CRNS placement was 2 to 5 times more likely to yield better SWE estimates compared to snow scales or currently available gridded products. CRNS provides valuable information about shallow, heterogeneous snowpacks in prairie and other environments and can benefit future missions from UAV and satellite platforms.
- Preprint
(7196 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 10 Mar 2025)
-
RC1: 'Comment on egusphere-2025-31', Markus Köhli, 02 Feb 2025
reply
Summary:
The authors H. Kim et al. describe in their article "Influence of Snow Spatial Variability on Cosmic Ray Neutron SWE" a very detailed assessment of CRNS snow water equivalent measurements at an agricultural site. The influence of heterogeneous snow cover on the CRNS signal is studied considering different sensor locations within their site. Finally, comparisons to other methods for snow water are provided. Currently, there is still a lack of CRNS studies focusing on snow cover. As it is a very complex topic, CRNS users are highly in need for such studies. Additionally, the comparison to remote sensing products is very interesting.General comments:
Article scope:
- The manuscript fits the scope of the journal.Quality:
- The manuscript is written in a straight forward way and can be easily followed. Some sentences might appear to be rather long and complex. The figure quality is good. The references are well organized.- One shortcoming of this study is that soil moisture was not consistently measured throughout the reference period. Sect. 3.2, citing Woodley et al., 2024, implies that in-situ samples were collected only once. With only considering snow heterogeneity the whole approach of the analysis of different influences on the CRNS signal has a synthetic character. The authors do not hide that fact, but they also do not clearly discuss it.
It would be completely out of scope to integrate soil moisture distributions here as well, yet, the findings from the authors may by such either be increased or smeared out. Soil moisture is most often correlated with the patches of snow, which means if areas with shallow snow pack additionally are more wet, the heterogeneity increases. On the other hand, all relative (SWE) signal variations depend on the underlying soil moisture. This can either decrease or increase the effect of snow. Such should clearly be discussed in order to provide a guide for non-CRNS experts for how to interpret the results.
- how is SWE calculated from the LIDAR measurements? Do the authors assume constant snow density?
- Is simplified weighting function (B1) from the appendix of DOI 10.5194/hess-21-5009-2017 not working for the SWE distribution (it does not have, to, the question is in fact more related to the Woodley publication).
- The comparison analysis is somehow confusing in the way it is structured. Like for soil moisture, you can assume, that there is a horizontal weighting function for SWE. There is no such of a systematic analysis existing in literature, the authors presented one solution in Woodley et al., 2024. The authors compare uniformly assumed SWE with the actual SWE distribution patters (and Eq.(1)). They compare equal weighting with simulations of detectors at 25 virtual locations. The authors find in their analysis that inhomogeneities in the SWE distribution lead to deviations from the assumed equal weighting. That part is discussed quantitatively in detail. They then switch from count rate considerations to spatial representativeness, where they compare a lidar SWE evaluation with an analytically weighted SWE. Here they find that the CRNS weighted SWE is more close to the area average.
Why are those results or evaluations not combined into one discussion? First, the neutron simulation also directly provides the neutron density over the study site. This density can be transformed to a SWE value can easily be compared to other ones. The virtual detectors you only need to describe where neutrons come from, to trace for example the origins individually. For just the count rate the entire detector layer in URANOS provides the values at each pixel in the domain. Secondly, after the probably a little bit too detailed quantitative discussion you repeat the comparison of SWE average vs. heterogeneous by only changing the averaging radius from 171 m to the whole domain. Whereas that is a part to be considered it takes a lot of room, especially as the results are not so different. Thirdly, there is switch from counts and comparison to averages to which points in their domain correspond to the average and therefore are more representative by now including another method which is analytical weighting.
In summary, that part should be condensed and restructured.Specific comments:
Comments to the figures:
- Labels A and B in the figures are referred to in the caption as (a) and (b).
- Caption Fig. 6: "surrounding the virtual" -> "surrounding the virtual detector".
- Fig. 7 seems to not add a lot more to what can already be seen in Fig. 6.
- Fig. 8: in a) and b) the dashed line could receive a label in order make more clear what you want to compare. The label "1 m" in the legend is not very descriptive.
- Fig. 9: convert the axis to normal dates. DOWY is used nowhere else in the manuscript.
- Fig. 9: The "CRNS timeseries" does not match anything which is plotted in the reference Woodley et al., 2024.
- Fig. 9: The authors include the "CRNS timeseries" - do they mean SWE derived from CRNS?Title:
- the reviewer thinks that the title "Influence of Snow Spatial Variability on Cosmic Ray Neutron SWE" overstates the results as the title is too general, although the research topic presented is a very broad basis for the SWE analysis. It should include that it primarily conceptualizes SWE analysis as derived from their site.Comments to the text body:
Introduction
- The overview is very well organized and includes the relevant literature. The different parts are, however, somehow disconnected. It begins with agricultural land-use changes, moves into dryland cropping techniques and then shifts to snow without clear linkage.
- l72: Hydrogen 'trap' (the technical term would be 'absorb') free neutrons if neutrons are thermalized. The CRNS signal attenuation is mainly due to slowing down of neutrons by hydrogen, i.e. losing energy through elastic collisions.Data and Methods
- The description of the NASA SnowEx field campaign at CARC is informative, but the connection between different measurement techniques (InSAR, UAV lidar, SfM, snow pits, etc.) is not well worked out given the specificities of each. Considering as the comparison of CRNS SWE with various gridded datasets (UCLA-re, SNODAS, and UA) is in general an important validation step and also undertaken in this study.Results and Discussion
- l198: "N_theta is the calibration neutron count, from the “snow-off” reference date of 15 January 2021." That is of course not a real calibration, it is an approximation for the calibration, which is associated with significant systematic uncertainties. One would need to for example look at the entire CRNS timeseries in order to try to understand (or justify) this choice. Even if you think, that this is a 'good enough' choice this needs to be discussed very clearly, as later on in the text you make very detailed numerical comparisons of which some directly relate to the choice of this value.
- l263: How significant are the relative differences mentioned here and shown in Fig. 5? The reviewer assumes, for example based on Fig. 6, that the number of neutrons counted rather low given the small relative differences. So that analysis might not be precise enough for sub-percent statements (i.e. 3.16%)?
- l288: As stated above, the overestimation might simply be a result of not using any weighting function, but there might other effects playing a role, too.
- l353: you mention 624 simulations but not how you arrive at that number, given that, the reader is left wondering what that means.Conclusion:
- l487: "a naively sited CRNS instrument (i.e., with no knowledge of the snow distribution) is still 2 to 5 times
more likely to be representative of the large-scale average SWE than a more conventional (...)" - that statement is not found in the discussion before.Technical comments:
Typography:
- Equations are part of the text body. Therefore, they follow the interpunctuation. That means a dot after Eq. (1), not before, and a comma after (2) and (3).
- Tables: Try using the same number of decimals and align numbers either left or, preferably, right.References:
- This manuscript refers to for the (previously already published) data. However, the DOI which is provided in Woodley et al., 2024 for the CRNS SWE data (https://doi.org/10.5067/NJR0AMMOHZ4E) does not work.Citation: https://doi.org/10.5194/egusphere-2025-31-RC1
Interactive computing environment
Data and Model Results for Spatial Analysis of CRNS at the CARC Haejo Kim and Sam Tuttle https://doi.org/10.5281/zenodo.14592408
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 70 | 23 | 5 | 98 | 2 | 2 |
- HTML: 70
- PDF: 23
- XML: 5
- Total: 98
- BibTeX: 2
- EndNote: 2
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1