the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Jul 2023
RoGeR v3.0.3 – a process-based hydrological toolbox model in Python
Abstract. Although water availability and water quality are equally important for an effective water resources management, to date, a combined representation of soil water balance components and water quality components in Python is not available. The new RoGeR toolbox contains models that can be used for the quantification of hydrological processes, fluxes and stores, but also solute transport processes based on StorAge selection. This study presents the code structure and functionalities of RoGeR developed as a scientific model toolbox following defined open-source software guidelines. RoGeR uses five different computational back ends covering just-in-time compilation, parallelism and graphical-processing units that might be used for optimizing computational performance. We show that graphical-processing unit computing have the greatest potential to improve computation time and energy usage especially for large modelling experiments. A simple modelling experiment highlights the capabilities of the new RoGeR model toolbox. We simulated the soil water balance, stable water isotope (18O) transport and corresponding travel time distributions through the soil of a grassland plot for a three-year period. Further development of RoGeR as a scientific software is possible and also needed due to the current limitations for a variety of process components and easily possible due to the open software architecture.
- Preprint
(3101 KB) - Metadata XML
-
Supplement
(3833 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on gmd-2023-118', Anonymous Referee #1, 18 Aug 2023
The authors present an interesting new hydrologic modelling toolbox in Python. The toolbox itself is very interesting and the publication is appropriate for GMD. I have a few points that the authors might want to address to improve their manuscript.
(1) The toolbox itself is interesting and is mostly well described. What took me a while to understands though, is whether the toolbox models only described 1-D soil processes or whether they can present a grid with spatial variability. As well as how this raster would be connected. My suggestion is that the authors expand Figure 1 to also include spatial and temporal discretization, as well the inputs to the model. I think it would help the reader to have such a conceptual figure regarding the possible model architectures.
(2) A related aspect is the very specific input requirements for the model discussed in lines 112ff. I am a bit confused why this is so specific when the model architecture is meant to be modular. Should the model not be adjustable to the data available? Precipitation of 10 minute time steps or less will limit available datasets very much. Why can the model not be run with hourly time steps? Especially given that other inputs are daily. I think the authors should provide some justification here, rather than just stating this as a fact.
(3) Another part where I would expect a bit more discussion is section 2.2. Much has been published on the issue of modular hydrologic models over the last 20 years, but hardly anything from this discussion is mentioned here (e.g. https://gmd.copernicus.org/articles/12/2463/2019/gmd-12-2463-2019.pdf). This discussion should also include how a certain process module (e.g. transpiration or surface runoff) could be represented using different process complexities. I assume this is included here.
(4) Could the authors expand a bit on the numerical implementation of the models and how this links to previous discussions on the topic (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2009WR008894)?
(5) In line 248, the authors mention that parameters are randomly generated, but do not explain more. I assume the authors sampled from independent uniform distributions. Are there other options than this approach? Here the link to other toolboxes might be particularly useful to further explore the model outputs given that more and more analysis algorithms can be run with generic input samples (e.g. https://www.sciencedirect.com/science/article/abs/pii/S1364815218303220).
(6) If the model is represented as a grid, is there a possibility for river flow actually occurring? Or is the domain limited to plot scales where channelized flow might not occur?
(7) A hydrologic modelling toolbox like the one presented here can (and probably should) link to a wider range of Python toolboxes that assess and attribute uncertainties (e.g. https://safetoolbox.github.io/) or estimate parameters (e.g. https://pymoo.org/). Discussing how one might use existing toolboxes with RoGeR would be interesting for users I believe.
(8) There are a few issues with the writing. E.g. "measurements for their states" should be measurements of, the last sentence in the abstracts includes and twice and is too long, ... So please have another read through just for grammar.
Citation: https://doi.org/10.5194/gmd-2023-118-RC1 -
AC2: 'Reply on RC1', Robin Schwemmle, 10 Jan 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2023-118/gmd-2023-118-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Robin Schwemmle, 10 Jan 2024
-
RC2: 'Comment on gmd-2023-118', Anonymous Referee #2, 17 Sep 2023
This paper presents a python library to implement RoGer hydrological model. The main reason behind developing this library was to provide an-easy-to-use, reproducible and modular code. The library itself can be quite useful given that it also tries to simulate water isotope data.
A major limitation of the paper is that it does not represent a real-world application of the code; we do not learn much by the synthetic example shown. The real challenge in implementing a distributed hydrological model is estimating the parameters that should be used for simulations, especially given the uncertainties in already limited hydrological data. Therefore, I suggest that the authors present a real-world case study in the paper. Also, a brief discussion of the calibration problem might benefit the paper; is there any plan to include calibration modules in the package?
My specific comments are in the attached PDF.
-
AC1: 'Reply on RC2', Robin Schwemmle, 10 Jan 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2023-118/gmd-2023-118-AC1-supplement.pdf
-
AC1: 'Reply on RC2', Robin Schwemmle, 10 Jan 2024
Status: closed
-
RC1: 'Comment on gmd-2023-118', Anonymous Referee #1, 18 Aug 2023
The authors present an interesting new hydrologic modelling toolbox in Python. The toolbox itself is very interesting and the publication is appropriate for GMD. I have a few points that the authors might want to address to improve their manuscript.
(1) The toolbox itself is interesting and is mostly well described. What took me a while to understands though, is whether the toolbox models only described 1-D soil processes or whether they can present a grid with spatial variability. As well as how this raster would be connected. My suggestion is that the authors expand Figure 1 to also include spatial and temporal discretization, as well the inputs to the model. I think it would help the reader to have such a conceptual figure regarding the possible model architectures.
(2) A related aspect is the very specific input requirements for the model discussed in lines 112ff. I am a bit confused why this is so specific when the model architecture is meant to be modular. Should the model not be adjustable to the data available? Precipitation of 10 minute time steps or less will limit available datasets very much. Why can the model not be run with hourly time steps? Especially given that other inputs are daily. I think the authors should provide some justification here, rather than just stating this as a fact.
(3) Another part where I would expect a bit more discussion is section 2.2. Much has been published on the issue of modular hydrologic models over the last 20 years, but hardly anything from this discussion is mentioned here (e.g. https://gmd.copernicus.org/articles/12/2463/2019/gmd-12-2463-2019.pdf). This discussion should also include how a certain process module (e.g. transpiration or surface runoff) could be represented using different process complexities. I assume this is included here.
(4) Could the authors expand a bit on the numerical implementation of the models and how this links to previous discussions on the topic (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2009WR008894)?
(5) In line 248, the authors mention that parameters are randomly generated, but do not explain more. I assume the authors sampled from independent uniform distributions. Are there other options than this approach? Here the link to other toolboxes might be particularly useful to further explore the model outputs given that more and more analysis algorithms can be run with generic input samples (e.g. https://www.sciencedirect.com/science/article/abs/pii/S1364815218303220).
(6) If the model is represented as a grid, is there a possibility for river flow actually occurring? Or is the domain limited to plot scales where channelized flow might not occur?
(7) A hydrologic modelling toolbox like the one presented here can (and probably should) link to a wider range of Python toolboxes that assess and attribute uncertainties (e.g. https://safetoolbox.github.io/) or estimate parameters (e.g. https://pymoo.org/). Discussing how one might use existing toolboxes with RoGeR would be interesting for users I believe.
(8) There are a few issues with the writing. E.g. "measurements for their states" should be measurements of, the last sentence in the abstracts includes and twice and is too long, ... So please have another read through just for grammar.
Citation: https://doi.org/10.5194/gmd-2023-118-RC1 -
AC2: 'Reply on RC1', Robin Schwemmle, 10 Jan 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2023-118/gmd-2023-118-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Robin Schwemmle, 10 Jan 2024
-
RC2: 'Comment on gmd-2023-118', Anonymous Referee #2, 17 Sep 2023
This paper presents a python library to implement RoGer hydrological model. The main reason behind developing this library was to provide an-easy-to-use, reproducible and modular code. The library itself can be quite useful given that it also tries to simulate water isotope data.
A major limitation of the paper is that it does not represent a real-world application of the code; we do not learn much by the synthetic example shown. The real challenge in implementing a distributed hydrological model is estimating the parameters that should be used for simulations, especially given the uncertainties in already limited hydrological data. Therefore, I suggest that the authors present a real-world case study in the paper. Also, a brief discussion of the calibration problem might benefit the paper; is there any plan to include calibration modules in the package?
My specific comments are in the attached PDF.
-
AC1: 'Reply on RC2', Robin Schwemmle, 10 Jan 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2023-118/gmd-2023-118-AC1-supplement.pdf
-
AC1: 'Reply on RC2', Robin Schwemmle, 10 Jan 2024
Model code and software
roger Robin Schwemmle https://github.com/Hydrology-IFH/roger
Calculating energy usage of RoGeR using Green Algorithms Robin Schwemmle https://doi.org/10.5281/zenodo.8095094
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 487 | 225 | 30 | 742 | 50 | 12 | 20 |
- HTML: 487
- PDF: 225
- XML: 30
- Total: 742
- Supplement: 50
- BibTeX: 12
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1