the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Apr 2024
A new dunetracking tool to support input parameter selection and uncertainty analyses using a Monte Carlo approach
Abstract. Precise and reliable information about bedforms, regarding geometry and dynamics, is relevant for many applications – like ensuring safe conditions for navigation along the waterways, parameterizing the roughness of the riverbed in numerical models, or improving bedload measurement and monitoring techniques. There are many so-called dunetracking tools to extract this information from bathymetrical data. However, most of these tools require the setting of various input parameters, which in turn influence the resulting bedform characteristics. How to set the values for these parameters and what influence they have on the calculations has not yet been comprehensively investigated. This is why we introduce a new dunetracking tool, which is able to quantify the influence of varying input parameter settings by performing a Monte Carlo Simulation. The core of the tool is a combination of the two existing applications Bedforms-ATM (Guitierrez, 2018) and RhenoBT (Frings et al., 2012), which have been extended by adding additional features. A wavelet analysis has been adapted from Bedforms-ATM while a zerocrossing procedure and a cross correlation analysis have been implemented based on RhenoBT. The combination of both tools enables a more accurate and sound procedure, as the results of the first step are required input parameters in the second step. By performing a Monte Carlo Simulation, comprehensive sensitivity analyses can be carried out and the possible range of results is revealed. At the same time, the high degree of automation allows the processing of large amounts of data. By applying the tool to a test dataset, it was found that bedform parameters react with different sensitivity to varying input parameter settings. Bedform lengths appeared to be more sensitive (uncertainties up to 50 % were identified) than bedform heights. The setting of a window size in the zerocrossing procedure (especially for the upper layer of bedforms in case secondary bedforms are present) was identified to be the most decisive input parameter. Here, however, the wavelet analysis offers orientation by providing a range of plausible input window sizes and thus allows a reduction of uncertainties. By choosing values outside this range, divergence behavior could be observed for several resulting bedform parameters. Concurrently, the time interval between two successive measurements has proven to have a significant influence on the determination of bedform dynamics. For the test dataset, the faster migrating secondary bedforms were no longer traceable for intervals longer than two hours. At the same time, they contributed to up to 90 % of the total bedload transport, highlighting the need for measurements in high temporal resolution in order to avoid a severe underestimation.
- Preprint
(3249 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-579', Anonymous Referee #1, 30 Apr 2024
The authors combined two existing morphometric approaches in a single dune tracking tool and applied it in a Monte Carlo Simulation to investigate the influence of varying input parameter settings and quantify the (un)certainty of outputs. The methodology and results of this sensitivity analysis are described in much detail and supported by multiple illustrations. I am convinced that the study is worth publishing, yet the manuscript is relatively long and mainly descriptive. This is also reflected in the discussion section which describes new data rather than interpret the existing results and discuss implications, e.g. recommendations for future morphometric studies. Overall, the analyses appear scientifically sound and relevant, but I have some suggestions on how to adjust the focus and improve transferability. Please find these suggestions as comments in the attached PDF file.
Furthermore, I provide my responses to the reviewing questions as a guideline here:
1. Does the paper address relevant scientific questions within the scope of ESurf?
The paper addresses relevant scientific questions that fit ESurf’s scope but its narrative and focus could be improved.2. Does the paper present novel concepts, ideas, tools, or data?
Although tools and concepts are taken from previous studies, the presented analysis is new and worth discussing.3. Are substantial conclusions reached?
The conclusions are relevant but mainly descriptive in their current form. The authors should try to integrate implications and recommendations for future studies.4. Are the scientific methods and assumptions valid and clearly outlined?
The methods and assumptions are suitable and described clearly.5. Are the results sufficient to support the interpretations and conclusions?
The results are well described and in much, if not too much, detail. But the manuscript falls short with respect to their interpretation and contextualization. As outlined before, this leads to conclusions which are mainly descriptive.6. Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?
The methodology is described in great detail and looks readily reproducible.7. Do the authors give proper credit to related work and clearly indicate their own new/original contribution?
The authors give proper credit to the original sources of their methodology but, in my view, should focus more on their own contribution which is the sensitivity analysis for an exemplary dune tracking algorithm (cf. next comment). Especially with respect to the current title, the authors fail to summarize the current state-of-the-art of bedform identification and previous studies on the subjectivity of their results.8. Does the title clearly reflect the contents of the paper?
This is my central point: the title implies that the authors present “a new dunetracking tool” which I think i) they don’t and ii) is not their main result. The authors even state that “There are many so-called dunetracking tools ...” So why present yet another one?
I would hence strongly recommend to clarify what is the novelty of this study, i.e. the quantification of uncertainty resulting from different (non-objective) input parameters and their weighting calculated for one exemplary dune tracking algorithm. Furthermore, this finding calls for recommendations to be followed in future morphometric analyses.9. Does the abstract provide a concise and complete summary?
The abstract is overall concise and complete (notwithstanding the previously raised issues).10. Is the overall presentation well-structured and clear?
The presentation of results is well-structured and clear, yet in my view too detailed and descriptive (the total number of figures is 18). The manuscript could dwell more on implications and recommendations and limit results to the essential.11. Is the language fluent and precise?
The language is fluent and, apart from very few occasions, precise.12. Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?
Mathematical formulae, symbols etc are used correctly.13. Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?
As outlined in the provided PDF, some parts of the paper could be reduced or combined, e.g. Figures 11/13/14. Others, such as the discussion chapter, could be extended.14. Are the number and quality of references appropriate?
Considering the research question of this study, I was surprised to see that only a few dune tracking tools were discussed and existing studies on their subjectivity were not references. I would suggest extending the literature research and improving the contextualization of results.15. Is the amount and quality of supplementary material appropriate?
Currently, the Appendix contains a table of measurement pairs which is needed as context. As outlined in the PDF comments, it could be helpful to be more explicit about the essential differences of these pairs, i.e. mainly the time shift between measurements. -
RC2: 'Comment on egusphere-2024-579', Anonymous Referee #2, 19 May 2024
The manuscript „A new dunetracking tool to support input parameter selection and uncertainty analyses using a Monte Carlo approach” deals with an interesting and relevant topic and is of interest for both the scientific community and practitioners. The proposed method to determine dune geometry and celerity is based on the combination of two approaches from the literature and is outlined in very much detail. However, it can, in my opinion, still be sharpened. The procedure is then applied to a field data set for which further details should be given. The obtained results are interesting and highlight the need for refined and harmonised methods to enable comparisons with results from other field sites (and hence studies). I do have some more comments on the manuscript (see below) so that I recommend returning the manuscript to the authors for revisions.
Detailed comments:
L13: There is novelty, but one may argue that it is an extended tool that is based on the combination of two previously published approaches (as mentioned a few lines below).
L17/18: It is not really clear what is meant by “first step” and “second step” - this becomes only clear after reading the manuscript.
L20: It could be mentioned that field data are used.
L23: Just a comment – this result is not really surprising and can be expected.
L27: See my last comment. In this context, since a specific time interval is given, I do recommend to indicate that field data collected at the river Rhine are analysed.
L43: Please indicate the label ‘a’ also in the references (L684/L686).
L44: The reference Henning (2013) is only cited with the title in German. Please provide more information where this was published.
L49: I fully agree and it is good to see that this is explicitly mentioned.
L57: I recommend deleting “highly”.
L61: (Guitierrez, 2018) must be (Guitierrez et al., 2018).
L65: “analog” or “analogue”?
L71: But the proposed procedure does not allow for an evaluation of 3D-dunes – correct?
L80: Personally, I don’t like starting a paragraph with a figure.
L85: Is this procedure also applicable in curved river sections or should it only be applied to straight sections?
L94: Strictly speaking, an analysis cannot “look at” something – I recommend rephrasing.
L104: What is the scale of the subsections? Multiple bedform lengths (and if so, how many)?
L105: It is a bit confusing that Figure 6 is already referenced here.
L106: Can you give some examples what kind of steps are meant here?
L107: What kind of geometric modeling?
L108: What about curved river sections (see my comment @L85) – how to generate meaningful BEPs for such cases?
L118: Please improve the caption of Figure 2, also by adding some more explanations (e.g., labels a/b should be used; which exemplary profile is shown (or is it a created one); why does x-axis start at 390 m; why is the minimum wavelength not limited to 0).
L128: How are these layers specified in the analysis (i.e., how to get the data for layer 1 / 2 from the total data set)?
L130: Can you please indicate how outliers were removed?
L133: Is it possible to define more than two layers? Hoe exactly are these layers determined? Some more guidance would be helpful for the reader.
L134: Isn’t this partly subjective contradicting the statement regarding automatization?
L135: I am not sure that I can follow this argumentation – I can three lines for the different significance levels and I see a total number of three detected peaks, but is not clear to me for which significance level (and how the different significance levels indicate a different number of layers). Some more explanations would be helpful here for the reader.
L138: See my comment @L134.
L154: What is the span of the moving average?
L157: Why does the lowest baseline separate the bedforms from the immobile bed? What if the bed below would also be moving with a constant speed (I know this is hypothetical, but so is the assumption that the bed below layer 2 is immobile)?
L159: “Several” indicates again that more than two layers can be defined (see my comment above).
Caption Figure 3: “For layer two the procedure is based on baseline 1 instead of the BEP.” This could be indicated in the text. It seems that a moving average was used for the layer two analysis (based on baseline 1 – what was the span)? How would one continue if there would be more than two layers?
On another note - which profile is shown in this plot (it seems to be different from the one shown in Figure 2)?L175: Figure 4: Why using again a different profile than the ones shown in Figure 3?
L176: There is no L_total defined in Figure 4.
L191: Check numbering of heading.
L200: See my comment @L65.
L206: See my comment @L43.
L214: What is a short measurement interval?
L216: There may be gaps in the data if too many outliers would be present, so that the migration rates may be biased?
L225: Does this mean bedforms corresponding to layer 2?
L228: Figure 5: Please define the data used for this plot (see my comments above) and a/b labels should be used. What was the time interval between the measurements?
L282: Fig.5 should be Fig. 6? Please be consistent in using Fig. or Figure in brackets (e.g., L306).
L288: Please provide more information on the data (especially spatial resolution; what kind of MBES was used and how was it deployed etc.).
L300: No. 5-14: are these the detailed measurements in Figure 6? How good was the spatial match of these profiles?
L306: See my comment before – what is the “raw resolution” of the data and what is the area of the MBS-footprint? What kind of interpolation algorithm was used to create the high-resolution grid with a point spacing of 10 cm? I assume that a total of four DEMs were derived (for No. 1-4 in Table 3)? How were the BEPs derived (there is some curvature of the channel)? How accurate were the
L306: What is meant by “normalized”? Is it meant that the mean bed elevation in this figure corresponds to zero? However, this value is different from the z-range shown in the previous figures. Please explain. Can you also comment more on the lateral differences in bed elevation profiles?
L307: In this context – in my opinion, the largest bedforms are present in BEP 1-4. Please provide some more explanations. Can you also please indicate BEPs 1-16 in Figure 6 (i.e. where is BEP 1 and where is BEP 16 – that will be helpful for the reader).
L330: It is interesting to note that the max. wavelength is about 25 m. Note that Lokin et al. (2022) (https://doi.org/10.1029/2021GL097127) identified wavelengths of up to 140 m in the downstream Waal river (although for lower discharges) – nonetheless, for higher flows (Q = 4000 m^3/s) the wavelengths were still > 60m. This is quite a difference that may be explained in the manuscript (I guess Lokin et al. 2022 should also be referenced as it deals with a similar topic – showing the need to adjust the literature review).
L336: I am not sure that the “sigma”-sign was appropriately defined in the text.
L344: Why reference to Figure 8 showing wavelengts? Should be Figure 9?
L349: If this statement refers to Figure 9 (what I assume), I cannot see these 10 m.
L394: Please give a reference for Lowess smoothing.
L431: A closing bracket is missing.
L459: “sizes” should be “size”
L460: Are BEPs 8-14 meant?
L462: Do you mean it is sensitive for window-sizes < 5 m?
L464: Is this because individual high bedforms are not as significant in the “T” analysis compared to the “H”-analysis?
L475: Are there other studies available that have reported bedform geometries in the Rhine river (in regions nearby) that can be used to validate the findings (see, e.g. Lokin 2022).
L499: See my comment @L475.
L500: Figure 16 should be Figure 17.
L534: Figure 18: Are there bed-load measurements available to verify the determined total transport rates?
L538: Section 5.2: In my opinion, this paragraph is not really a discussion. It is rather a summary and repetition of the contents described before.
L547: delete “shortly”.
L551: “repeat” should be “repeated”.
Citation: https://doi.org/10.5194/egusphere-2024-579-RC2 -
AC1: 'Comment on egusphere-2024-579', Julius Reich, 27 May 2024
Many thanks to the referees for the interest in our work and the detailed feedback. We are convinced that the comments are of great value to further improve the manuscript and we will revise it accordingly.
We will pay particular attention to sharpening the focus of the paper and also reflect this in the title. We agree that not all of the individual parts of the method are entirely new, but the implementation of a highly automated Monte Carlo Simulation is an important enhancement, which is the main focus of our work.
Further on, we will try to reduce the method and results section more to the essentials and at the same time answer the outstanding questions. We will also provide some more information on the test dataset (in a separate appendix, if necessary). However, a detailed description of the hydrographic processing (such as geometric modelling) is not intended to be part of this manuscript, as it can be found in Lorenz et al. (2021) (https://doi.org/10.1007/s41064-021-00145-0), as referenced in the method section.
We will revise the discussion section based on the provided comments. From our point of view, the purpose of the manuscript is the introduction of the method (carrying out comprehensive sensitivity analyses using a Monte Carlo approach). Based on a test dataset we show the relevance of input parameter selection. In future morphometric analyses, sensitivity and uncertainty can be analyzed and compared for different bathymetries and different hydrological conditions, for which this manuscript can be used as a reference. For this purpose, the recently published benchmarking dataset included in Scheiber et al. (2024) could be considered. However, the implementation of such a comparative analysis is not the aim of this manuscript.
Citation: https://doi.org/10.5194/egusphere-2024-579-AC1
Model code and software
Software code (incl. test dataset) Julius Reich and Axel Winterscheid https://github.com/JRbfg/DTMCS
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 228 | 54 | 27 | 309 | 23 | 18 |
- HTML: 228
- PDF: 54
- XML: 27
- Total: 309
- BibTeX: 23
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1