Technical note: Seamless Extraction and Analysis of River Networks in R

Carraro, Luca

doi:https://doi.org/10.5194/egusphere-2023-939

Preprints

https://doi.org/10.5194/egusphere-2023-939

Preprints

10 May 2023

| 10 May 2023

Technical note: Seamless Extraction and Analysis of River Networks in R

Luca Carraro

Abstract. Spatially explicit mathematical models are key to a mechanistic understanding of environmental processes in rivers. Such models necessitate extended information on networks' morphology, which is often retrieved from geographic information system (GIS) software, thus hindering the establishment of replicable, script-based workflows. Here I present rivnet, an R-package for GIS-free extraction and analysis of river networks based on digital elevation models (DEMs). The package exploits TauDEM's flow direction algorithm on user-provided or online accessible DEMs, and allows computing covariate values and assigning hydraulic variables across any network node. The package is designed so as to require minimal user input, while allowing customization for experienced users. It is specifically intended for application in models of ecohydrological, ecological or biogeochemical processes in rivers. As such, rivnet aims to make river network analysis accessible to users unfamiliar with GIS-based and geomorphological methods, and therefore enhance the use of spatially explicit models in rivers.

Received: 08 May 2023 – Discussion started: 10 May 2023

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 6583 KB)

Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Preprint (6583 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

23 Oct 2023

Technical note: Seamless extraction and analysis of river networks in R

Luca Carraro

Hydrol. Earth Syst. Sci., 27, 3733–3742, https://doi.org/10.5194/hess-27-3733-2023,https://doi.org/10.5194/hess-27-3733-2023, 2023

Short summary

Luca Carraro

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2023-939', Polina Lemenkova, 24 May 2023

Review of the manuscript ‘Technical note: Seamless extraction and analysis of river networks in R’ by Carraro submitted to HESS.
Recommendation: ACCEPT.
Focus of the paper: the author presented a novel package “rivnet”, aimed at hydrological analysis and modelling river networks using digital elevation models (DEMs). The author developed his own novel package ‘rivnet’. Therefore, the paper presents a valuable contribution to the scientific methods development in Geosciences with a special focus on hydrological modelling.
Relevance: The presented study is the original primary research within scope of the journal Hydrology and Earth System Sciences. The author developed a new package which focuses on modelling of the environmental processes in rivers. The manuscript meets general criteria of the significance in applied programming (R), environmental data modelling and hydrology (analysis of catchment, watershed basin analysis, etc). It is relevant to the journal topic as corresponding to the major domain and research disciplines: hydrological modelling, Earth sciences, applied programming.
Abstract is well written, concisely and clearly describes the undertaken study.
Structure: The article is well organized with 4 structured sections. 1 Introduction (a background on hydrological modelling is provided); 2 Package overview (here, the author presents the package in details and well explains technical nuances); 3 Applications (two case studies); 4 Perspectives and limitations.
Logic: The clarity of the text logic and organization of the paper is sufficient with details on technical description of the package and theoretical basis of hydrological modelling. The paper consistently interprets the results with detailed explanations and comments on new package ‘rivnet’. A brief comparison of the package with those in previous approaches in hydrological modelling is presented.
Introduction presents a background, defines research goals and provides a clear statement of research problem. The Introduction well describes the research. Introduction and background show context of the article. Literature is well referenced and relevant.
Study area: is described with sufficient details.
Research questions and goal are identified: the author developed a novel package based on R syntax which aims at mathematical modelling of the hydrological processes. Objectives are relevant to the study aim with regard to river modelling (two cases of river modelling in study area of Switzerland).
Literature regarding the relevant topics is reviewed, formatted according to the journal rules and appropriately referenced. Evaluation diverse hydrological modelling aspects. Examples include papers on flow directions, upslope areas in grid DEMs, modelling river basins by fractals, GIS, modelling watersheds, geomorphic feature extraction from LiDAR etc. The literature is relevant and well selected.
Research gaps and weakness in former works are described lack of script-based hydrological package which would be used instead of GIS with extended functionality using R language.
Motivation is explained: this study contributes to fill in the gaps in the existing methods on river networks’ morphology modelling using replicable, script-based workflows. The functionality of the presented package ‘rivnet’ is well explained with provided details on commands.
English language: fine.
Data used in this study are described: To illustrate the functionality of package, the author presented two cases: 1) evaluation of elongation and aggregation properties of river networks based on data on main river stem in a catchment scales. This well demonstrated the computational performance of ‘rivnet’ package; 2) modelling phosphorous dynamics based on data regarding selected Swiss rivers; Here the author demonstrated how the package can calculate covariate values, assign hydraulic variables and pinpoint sites on the river network. Both case studies are well explained and illustrated.
Methods: Methods described with sufficient detail and information. The author explained the main functionality of the developed R package: it exploits the DEM’s flow direction algorithm based on DEMs, computes covariate values and assigns hydraulic variables across network nodes. The workflow is well structured and clearly described with sufficient information to reproduce the approach.
Results are reported: The main results consists in the novel developed package - the 'rivnet'. The author well explained the functionality, demonstrated its syntax and presented 2 case studies. The package enables the analysis of river networks in the R language. The emphasis is placed on environmental modelling and hydrological data analysis (and DEM). The package allows computation of areas and compute covariate values or hydraulic variables across the river network. Extraction of drainage patterns is based on TauDEM’s D8 flow direction algorithm. The author well explained and commented on the new developed package ‘rivnet’.
Discussion interpreted the major outcomes of this study. The advantages of the obtained results are described and compared with other studies. The Discussion described the issues of methodology and results.
Conclusion Conclusions are well stated: the ‘rivnet’ aims to make river network analysis supported by geomorphological modelling and the use of DEM. The Conclusions are linked to original research question, explain and support the results (developed package) and summarized the study. The conclusions are appropriately stated and connected to the original research goal of development of the hydrological package.
Actuality, novelty and importance of the research is clear. It consists in the novel developed package based on R which aims at hydrological modelling. The package is flexible enough: it requires minimal user input, and allows customization for experienced users.
Academic contribution: Rigorous investigation performed to a high technical and professional standard in software development in geosciences. The author presented a package which is intended for application in models of ecohydrological, ecological or biogeochemical processes in rivers. In this way, the novel package is multi-purpose and has potentially diverse applications. The new R package “rivnet” will support hydrological analysis, spatial modelling and investigation of geomorphic and environmental processes. Such study well deserved to be published in HESS.
Figures The authors presented 5 figures which well illustrate the functionality of the novel package. The figures are of fine quality, the illustrations are well done with carefully selected type of lines and colour palettes, aesthetically plotted and professionally made from the cartographic perspective. The figures are easy to read for hydrologists and geographers, relevant and suitable to this paper. Figures are labelled and appropriately described. They illustrate the applications of the new ‘rivnet’ package.
Recommendation: This manuscript can be ACCEPTED based on the detailed report above.
With kind regards,
- Polina Lemenkova.
24.05.2023.

Citation: https://doi.org/10.5194/egusphere-2023-939-RC1
- AC1: 'Reply on RC1', Luca Carraro, 07 Aug 2023
  
  Thank you very much for your very positive assessment of this work. As it seems, no replies to specific comments or manuscript revisions are requested.
  
  Citation: https://doi.org/10.5194/egusphere-2023-939-AC1
RC2:
'Comment on egusphere-2023-939', Warrick Dawes, 07 Aug 2023

egusphere-2023-939 “Technical note: Seamless extraction and analysis of river networks in R” Luca Carraro
This technical note details a workflow built on previous work, that allows a more or less transparent application of terrain analysis tools resulting in suitable outputs for river and landscape modellers. It appears to be very useful, and most important is coded in the publicly available environment R, with open source on CRAN.
The article itself is very well balanced, between background, description of the algorithm, and two useful applications. The English expression is good in both the technical and prosaic portions of text.
Better colour choices, or perhaps different line textures, would make Figure 1 better, as subtle changes in colour with very narrow lines makes the differentiation between “real river” and “real river & OCN” almost impossible.
Figure 2A is very hard to see with the streamlines only 1 pixel wide on a sea of purple and the magenta catchment outline. A different background colour would be better, or thicker lines for the streams. I think this is the Thur sub-catchment and it could be indicated as such in the caption. Also, the site figure hidden in Figure 3B deserves its own figure (before Fig 2?), with a few cities and the neighbouring countries indicated, so the rest seems more connected. Figure 5 does a good job of showing some plot() functionality as well as being a decent figure overall.
The D8 method is ideal for delineating topographic boundaries and reliably finding lowest elevation paths coincident with streams. Flow accumulation and paths may on the other hand be better described by D-inf methods partitioning flow to all lower elevation neighbours or to those along greatest local slope direction. Is there the potential within the workflow to delineate ridge and streamlines with D8 then impose those as constraints to estimate other parameters with alternate algorithms? This would not be important if simply using fractional areas to represent pollutant load or runoff potential, for example, but would be when considering flow paths across the surface.

Citation: https://doi.org/10.5194/egusphere-2023-939-RC2
- AC2: 'Reply on RC2', Luca Carraro, 07 Aug 2023
  
  Thank you very much for your very positive assessment of this work. Specific comments are hereafter addressed.
  Fig. 1. I agree that the distinction between arrow types in Fig. 1b may be unclear. In the revised manuscript, I will use dashed arrows for category "real river & OCN", hence making the distinction clearer. Thanks for spotting this.
  Fig. 2. The river structure in Fig. 2a is indeed hardly visible, which is due to each pixel being assigned a color, as opposed to the catchment outline, which is a polygon and can thus be plotted with increased line width. This is unfortunately unavoidable, because this figure appears when calling extract_river(), that is before streams have been delimited (which actually occurs when the subsequent function aggregate_river() is called). I already tried alternative color palettes, which however did not improve the quality of the figure. In the revised manuscript, I propose to use a coarser DEM resolution, thus resulting in larger pixels and hence better readability of the figure. By the way, the catchment shown in Fig. 2a is river Ilfis, as already mentioned in the caption.
  
  Fig. 3b. Thank you for this suggestion. However, I do not think that the inset figure (i.e., the map of Switzerland) should be a figure in itself, as this does not show an application of the use of rivnet, but only supports the two applications shown. However, I agree that some more details could be added to the inset. In the revised manuscript, I propose to place the y-axis of Fig. 3b on the right-hand side, thus making space for a bigger inset, in which e.g., borders of neighbouring countries, and meridians and parallels can be added.
  Use of D8 in combination with Dinf. Thank you very much for this insightful comment. It is indeed possible to extract catchment features obtained via the D8 method and apply them in combination with the Dinf method (or alternative algorithms) for more detailed catchment analyses. Indeed, the application of the Dinf method from the TauDEM library in R is allowed by the general-purpose function taudem_exec() of package traudem, while streamlines derived via the D8 method are produced as output of rivnet's aggregate_river(). Nonetheless, I deem that building a dedicated workflow for this option in rivnet would be out of scope with respect to the main focus of the package, which is supporting ecohydrological, ecological and biogeochemical models at catchment scale. However, in the revised manuscript, I will mention in the concluding section that it is possible to export river objects in free format and use them in subsequent landscape analyses with the Dinf or other methods.
  
  Citation: https://doi.org/10.5194/egusphere-2023-939-AC2

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2023-939', Polina Lemenkova, 24 May 2023

Review of the manuscript ‘Technical note: Seamless extraction and analysis of river networks in R’ by Carraro submitted to HESS.
Recommendation: ACCEPT.
Focus of the paper: the author presented a novel package “rivnet”, aimed at hydrological analysis and modelling river networks using digital elevation models (DEMs). The author developed his own novel package ‘rivnet’. Therefore, the paper presents a valuable contribution to the scientific methods development in Geosciences with a special focus on hydrological modelling.
Relevance: The presented study is the original primary research within scope of the journal Hydrology and Earth System Sciences. The author developed a new package which focuses on modelling of the environmental processes in rivers. The manuscript meets general criteria of the significance in applied programming (R), environmental data modelling and hydrology (analysis of catchment, watershed basin analysis, etc). It is relevant to the journal topic as corresponding to the major domain and research disciplines: hydrological modelling, Earth sciences, applied programming.
Abstract is well written, concisely and clearly describes the undertaken study.
Structure: The article is well organized with 4 structured sections. 1 Introduction (a background on hydrological modelling is provided); 2 Package overview (here, the author presents the package in details and well explains technical nuances); 3 Applications (two case studies); 4 Perspectives and limitations.
Logic: The clarity of the text logic and organization of the paper is sufficient with details on technical description of the package and theoretical basis of hydrological modelling. The paper consistently interprets the results with detailed explanations and comments on new package ‘rivnet’. A brief comparison of the package with those in previous approaches in hydrological modelling is presented.
Introduction presents a background, defines research goals and provides a clear statement of research problem. The Introduction well describes the research. Introduction and background show context of the article. Literature is well referenced and relevant.
Study area: is described with sufficient details.
Research questions and goal are identified: the author developed a novel package based on R syntax which aims at mathematical modelling of the hydrological processes. Objectives are relevant to the study aim with regard to river modelling (two cases of river modelling in study area of Switzerland).
Literature regarding the relevant topics is reviewed, formatted according to the journal rules and appropriately referenced. Evaluation diverse hydrological modelling aspects. Examples include papers on flow directions, upslope areas in grid DEMs, modelling river basins by fractals, GIS, modelling watersheds, geomorphic feature extraction from LiDAR etc. The literature is relevant and well selected.
Research gaps and weakness in former works are described lack of script-based hydrological package which would be used instead of GIS with extended functionality using R language.
Motivation is explained: this study contributes to fill in the gaps in the existing methods on river networks’ morphology modelling using replicable, script-based workflows. The functionality of the presented package ‘rivnet’ is well explained with provided details on commands.
English language: fine.
Data used in this study are described: To illustrate the functionality of package, the author presented two cases: 1) evaluation of elongation and aggregation properties of river networks based on data on main river stem in a catchment scales. This well demonstrated the computational performance of ‘rivnet’ package; 2) modelling phosphorous dynamics based on data regarding selected Swiss rivers; Here the author demonstrated how the package can calculate covariate values, assign hydraulic variables and pinpoint sites on the river network. Both case studies are well explained and illustrated.
Methods: Methods described with sufficient detail and information. The author explained the main functionality of the developed R package: it exploits the DEM’s flow direction algorithm based on DEMs, computes covariate values and assigns hydraulic variables across network nodes. The workflow is well structured and clearly described with sufficient information to reproduce the approach.
Results are reported: The main results consists in the novel developed package - the 'rivnet'. The author well explained the functionality, demonstrated its syntax and presented 2 case studies. The package enables the analysis of river networks in the R language. The emphasis is placed on environmental modelling and hydrological data analysis (and DEM). The package allows computation of areas and compute covariate values or hydraulic variables across the river network. Extraction of drainage patterns is based on TauDEM’s D8 flow direction algorithm. The author well explained and commented on the new developed package ‘rivnet’.
Discussion interpreted the major outcomes of this study. The advantages of the obtained results are described and compared with other studies. The Discussion described the issues of methodology and results.
Conclusion Conclusions are well stated: the ‘rivnet’ aims to make river network analysis supported by geomorphological modelling and the use of DEM. The Conclusions are linked to original research question, explain and support the results (developed package) and summarized the study. The conclusions are appropriately stated and connected to the original research goal of development of the hydrological package.
Actuality, novelty and importance of the research is clear. It consists in the novel developed package based on R which aims at hydrological modelling. The package is flexible enough: it requires minimal user input, and allows customization for experienced users.
Academic contribution: Rigorous investigation performed to a high technical and professional standard in software development in geosciences. The author presented a package which is intended for application in models of ecohydrological, ecological or biogeochemical processes in rivers. In this way, the novel package is multi-purpose and has potentially diverse applications. The new R package “rivnet” will support hydrological analysis, spatial modelling and investigation of geomorphic and environmental processes. Such study well deserved to be published in HESS.
Figures The authors presented 5 figures which well illustrate the functionality of the novel package. The figures are of fine quality, the illustrations are well done with carefully selected type of lines and colour palettes, aesthetically plotted and professionally made from the cartographic perspective. The figures are easy to read for hydrologists and geographers, relevant and suitable to this paper. Figures are labelled and appropriately described. They illustrate the applications of the new ‘rivnet’ package.
Recommendation: This manuscript can be ACCEPTED based on the detailed report above.
With kind regards,
- Polina Lemenkova.
24.05.2023.

Citation: https://doi.org/10.5194/egusphere-2023-939-RC1
- AC1: 'Reply on RC1', Luca Carraro, 07 Aug 2023
  
  Thank you very much for your very positive assessment of this work. As it seems, no replies to specific comments or manuscript revisions are requested.
  
  Citation: https://doi.org/10.5194/egusphere-2023-939-AC1
RC2:
'Comment on egusphere-2023-939', Warrick Dawes, 07 Aug 2023

egusphere-2023-939 “Technical note: Seamless extraction and analysis of river networks in R” Luca Carraro
This technical note details a workflow built on previous work, that allows a more or less transparent application of terrain analysis tools resulting in suitable outputs for river and landscape modellers. It appears to be very useful, and most important is coded in the publicly available environment R, with open source on CRAN.
The article itself is very well balanced, between background, description of the algorithm, and two useful applications. The English expression is good in both the technical and prosaic portions of text.
Better colour choices, or perhaps different line textures, would make Figure 1 better, as subtle changes in colour with very narrow lines makes the differentiation between “real river” and “real river & OCN” almost impossible.
Figure 2A is very hard to see with the streamlines only 1 pixel wide on a sea of purple and the magenta catchment outline. A different background colour would be better, or thicker lines for the streams. I think this is the Thur sub-catchment and it could be indicated as such in the caption. Also, the site figure hidden in Figure 3B deserves its own figure (before Fig 2?), with a few cities and the neighbouring countries indicated, so the rest seems more connected. Figure 5 does a good job of showing some plot() functionality as well as being a decent figure overall.
The D8 method is ideal for delineating topographic boundaries and reliably finding lowest elevation paths coincident with streams. Flow accumulation and paths may on the other hand be better described by D-inf methods partitioning flow to all lower elevation neighbours or to those along greatest local slope direction. Is there the potential within the workflow to delineate ridge and streamlines with D8 then impose those as constraints to estimate other parameters with alternate algorithms? This would not be important if simply using fractional areas to represent pollutant load or runoff potential, for example, but would be when considering flow paths across the surface.

Citation: https://doi.org/10.5194/egusphere-2023-939-RC2
- AC2: 'Reply on RC2', Luca Carraro, 07 Aug 2023
  
  Thank you very much for your very positive assessment of this work. Specific comments are hereafter addressed.
  Fig. 1. I agree that the distinction between arrow types in Fig. 1b may be unclear. In the revised manuscript, I will use dashed arrows for category "real river & OCN", hence making the distinction clearer. Thanks for spotting this.
  Fig. 2. The river structure in Fig. 2a is indeed hardly visible, which is due to each pixel being assigned a color, as opposed to the catchment outline, which is a polygon and can thus be plotted with increased line width. This is unfortunately unavoidable, because this figure appears when calling extract_river(), that is before streams have been delimited (which actually occurs when the subsequent function aggregate_river() is called). I already tried alternative color palettes, which however did not improve the quality of the figure. In the revised manuscript, I propose to use a coarser DEM resolution, thus resulting in larger pixels and hence better readability of the figure. By the way, the catchment shown in Fig. 2a is river Ilfis, as already mentioned in the caption.
  
  Fig. 3b. Thank you for this suggestion. However, I do not think that the inset figure (i.e., the map of Switzerland) should be a figure in itself, as this does not show an application of the use of rivnet, but only supports the two applications shown. However, I agree that some more details could be added to the inset. In the revised manuscript, I propose to place the y-axis of Fig. 3b on the right-hand side, thus making space for a bigger inset, in which e.g., borders of neighbouring countries, and meridians and parallels can be added.
  Use of D8 in combination with Dinf. Thank you very much for this insightful comment. It is indeed possible to extract catchment features obtained via the D8 method and apply them in combination with the Dinf method (or alternative algorithms) for more detailed catchment analyses. Indeed, the application of the Dinf method from the TauDEM library in R is allowed by the general-purpose function taudem_exec() of package traudem, while streamlines derived via the D8 method are produced as output of rivnet's aggregate_river(). Nonetheless, I deem that building a dedicated workflow for this option in rivnet would be out of scope with respect to the main focus of the package, which is supporting ecohydrological, ecological and biogeochemical models at catchment scale. However, in the revised manuscript, I will mention in the concluding section that it is possible to export river objects in free format and use them in subsequent landscape analyses with the Dinf or other methods.
  
  Citation: https://doi.org/10.5194/egusphere-2023-939-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Publish subject to minor revisions (further review by editor) (21 Aug 2023) by Roger Moussa

AR by Luca Carraro on behalf of the Authors (24 Aug 2023) Author's response Author's tracked changes Manuscript

ED: Publish as is (20 Sep 2023) by Roger Moussa

AR by Luca Carraro on behalf of the Authors (20 Sep 2023)

Journal article(s) based on this preprint

23 Oct 2023

Technical note: Seamless extraction and analysis of river networks in R

Luca Carraro

Hydrol. Earth Syst. Sci., 27, 3733–3742, https://doi.org/10.5194/hess-27-3733-2023,https://doi.org/10.5194/hess-27-3733-2023, 2023

Short summary

Luca Carraro

Interactive computing environment

rivnet: Extract and Analyze Rivers from Elevation Data Luca Carraro https://cran.r-project.org/web/packages/rivnet/index.html

Luca Carraro

Viewed

Total article views: 576 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
391	168	17	576	13	12

HTML: 391
PDF: 168
XML: 17
Total: 576
BibTeX: 13
EndNote: 12

Views and downloads (calculated since 10 May 2023)

Month	HTML	PDF	XML	Total
May 2023	156	66	4	226
Jun 2023	52	19	1	72
Jul 2023	52	21	2	75
Aug 2023	81	25	5	111
Sep 2023	34	27	2	63
Oct 2023	16	10	3	29
Nov 2023	0
Dec 2023	0
Jan 2024	0
Feb 2024	0
Mar 2024	0
Apr 2024	0
May 2024	0
Jun 2024	0
Jul 2024	0
Aug 2024	0
Sep 2024	0

Cumulative views and downloads (calculated since 10 May 2023)

Month	HTML	PDF	XML	Total
May 2023	156	66	4	226
Jun 2023	52	19	1	72
Jul 2023	52	21	2	75
Aug 2023	81	25	5	111
Sep 2023	34	27	2	63
Oct 2023	16	10	3	29
Nov 2023	0
Dec 2023	0
Jan 2024	0
Feb 2024	0
Mar 2024	0
Apr 2024	0
May 2024	0
Jun 2024	0
Jul 2024	0
Aug 2024	0
Sep 2024	0

Viewed (geographical distribution)

Total article views: 575 (including HTML, PDF, and XML) Thereof 575 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 04 Sep 2024

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (6583 KB)
Metadata XML

Short summary

Mathematical models are key to the study of environmental processes in rivers. Such models often require information on river morphology from geographic information system (GIS) software, which hinders the use of replicable workflows. Here I present rivnet, an R package for simple, robust, GIS-free extraction and analysis of river networks. The package is designed so as to require minimal user input and is oriented towards ecohydrological, ecological and biogeochemical modelling.


Total:	0
HTML:	0
PDF:	0
XML:	0