SWAT+MODFLOW: A New Hydrologic Model for Simulating Surface-Subsurface Flow in Managed Watersheds

Bailey, Ryan; Abbas, Salam; Arnold, Jeffrey; White, Michael

doi:https://doi.org/10.5194/egusphere-2025-300

Preprints

https://doi.org/10.5194/egusphere-2025-300

Preprints

20 Mar 2025

| 20 Mar 2025

SWAT+MODFLOW: A New Hydrologic Model for Simulating Surface-Subsurface Flow in Managed Watersheds

Ryan Bailey, Salam Abbas, Jeffrey Arnold, and Michael White

Abstract. Coupled surface-subsurface hydrologic models are used worldwide to study historical patterns of water storage and hydrologic behaviour, investigate the impact of management strategies on water resources, and quantify the impact of changing climate, population, and policies. This study presents a new hydrologic model to simulate surface and subsurface in a physically based spatially distributed manner by linking the popular SWAT+ and MODFLOW modelling codes. Within this new code, SWAT+ simulates processes in the landscape, soils, channels, and reservoirs, whereas MODFLOW simulation groundwater processes and interaction with land surface features (soil, channels, canals, reservoirs, tile drains). Geographic connections between SWAT+ objects and MODFLOW grid cells are established a priori using a GIS and then read into the code to be used throughout the simulation to map hydrologic fluxes (recharge, soil water transfer, groundwater-channel exchange, canal seepage, tile drainage outflow, groundwater-reservoir exchange, pumping for irrigation) on a daily time step. The use and general accuracy of the model is demonstrated for two study regions that are subject to irrigation management: the Arkansas River Basin in Colorado and the San Joaquin River Basin in California. An accompanying tutorial and example model data allow for easy use of the model to other study regions. As both SWAT+ and MODFLOW are widely used worldwide for watershed and groundwater modelling, we expect that this new tool can be an important asset in many water resources projects.

Received: 21 Jan 2025 – Discussion started: 20 Mar 2025

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 paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 3025 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 (3025 KB)

Supplement (3333 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

05 Sep 2025

SWAT+MODFLOW: a new hydrologic model for simulating surface-subsurface flow in managed watersheds

Ryan T. Bailey, Salam Abbas, Jeffrey G. Arnold, and Michael J. White

Geosci. Model Dev., 18, 5681–5697, https://doi.org/10.5194/gmd-18-5681-2025,https://doi.org/10.5194/gmd-18-5681-2025, 2025

Short summary

Ryan Bailey, Salam Abbas, Jeffrey Arnold, and Michael White

Interactive discussion

Status: closed

CC1: 'Comment on egusphere-2025-300', Nima Zafarmomen, 02 Apr 2025

This manuscript presents a technically robust and well-structured integration of SWAT+ with MODFLOW-NWT, offering a significant modeling advancement for coupled surface-subsurface hydrology in managed watersheds.
The authors clearly state that the purpose is to introduce the model, not to fully calibrate it. However, for a model intended to support decision-making in complex, managed basins, stronger evidence of predictive accuracy (e.g., calibrated groundwater heads, fluxes, or uncertainty estimates) would reinforce its credibility.
NSE values are not reported for groundwater head simulations. Including these would help assess model robustness.
The use of a transfer function to represent vadose zone processes is a limitation. The authors should elaborate more on how this simplification affects recharge timing and suggest pathways for integrating UZF in future versions.

While MODFLOW can simulate confined aquifers, the manuscript uses only unconfined systems. Given the relevance of confined systems in agricultural basins, this limits the model's broader applicability and should be more explicitly acknowledged.
The poor performance (NSE = –0.31) in upstream segments of the San Joaquin River is concerning. Although this is not the paper’s central focus, the authors should discuss potential reasons (e.g., inflow data, canal routing) and whether additional calibration steps might help.
While the model runs are reasonably fast, the authors could include more benchmarks or comparisons with SWAT+ alone to quantify the cost-benefit of integrating MODFLOW, especially for large-scale applications.

Citation: https://doi.org/10.5194/egusphere-2025-300-CC1
RC1:
'Comment on egusphere-2025-300', Gerrit H. de Rooij, 07 Apr 2025

Dear editor, dear authors,
I reviewed the paper in detail, and the tutorial cursorily. The case for the model introduced by the authors is made convincingly, and the model clearly falls within the scope of GMD.

Although the model consists of existing components that have been augmented and linked, there is enough new material in the paper to justify publication in GMD. The relevance for practical applications is explained well by the authors.

The tutorial could use a few more references (including links) to established methods/techniques/software that the authors used. Examples of established software used are Notepad++ and QGIS, which are given without link or reference. Several links to other useful resources are given in the text of the tutorial, but not included in the reference list, which is a bit inconvenient.

The tutorial explains the working principle of PEST very well, which was quite helpful. But the Morris method used for the global sensitivity analysis used by PEST++ appears without explanation or reference.

All in all, I have no major comments. A few unclarities and inconsistencies appear in the text, highlighted in the minor comments directly inserted into the text. I suspect the inconsistencies are textual, rather than substantial, and can therefore be removed by a suitable rewriting of the text. I do not think the model code itself needs corrections. If this assessment proves to be correct, minor revisions should be sufficient.

Sincerely,
Gerrit de Rooij

Citation: https://doi.org/10.5194/egusphere-2025-300-RC1
- AC1: 'Reply on RC1', Ryan Bailey, 02 Jun 2025
  
  We have replied to all comments from RC1 in the attached document.
  Black text = reviewer text
  Red text = author response
  Blue text = text from the manuscript, used in author response.
  
  Citation: https://doi.org/10.5194/egusphere-2025-300-AC1
RC2:
'Comment on egusphere-2025-300', Kerry Callaghan, 22 Apr 2025

This paper presents a new coupling of two existing models, SWAT+ (watershed hydrology) and MODFLOW-NWT (groundwater flow). The coupled model is more hydrologically complete than either of the models on their own, and would be a useful tool for many regional hydrological studies. The paper falls within the scope of GMD, and I believe that it will be fit for publication after minor revisions have been completed. I think that this paper and the tool it describes are useful, but certain explanations need to be clarified and figures in particular need some work to aid the reader’s understanding.
This paper provides an updated coupling after a 2020 paper (referenced Bailey et al., 2020a in the paper) already linked SWAT+ and MODFLOW. The motivation given for the new paper is that SWAT+ has been modified extensively over the past 5 years. One question that I have is whether the new linkage will be able to absorb future improvements to SWAT+, or will brand new couplings be required every few years as modifications continue?
The paper links to a tutorial document, executable file, and sample input data to attempt reconstructions of the simulations presented. I appreciate this effort and the inclusion of the sample data. However, as a Linux user, I was unable to run the .exe file. Source code is provided; it would be extremely helpful to include compilation instructions.
Line 90: The paragraph starting at this line seems repetitive and unnecessary. Can it be integrated into the above paragraph?
Figure 1: This This figure needs some work, e.g. both the Arkansas River and the San Joaquin River have the same symbology, so maybe it would work better to use the same symbol and then just specify on which panel does it refer to which river. What are the small black dots on panel B? Why does panel C have the additional aquifer information while panel A does not?
Line 163 states that for the JMR, the aquifer boundary is the same as the watershed boundary (Fig 1B), but Fig 1B does not actually show this – it would be clearer if explicitly shown or mentioned in the caption.
Are aquifer boundaries shown in this figure and mentioned in the text from an external dataset? Table 2 includes aquifer thickness but it is not clear if this dataset also defines the boundaries.
Line 173 begins a description of population grid cell properties for the MSJ-LC model. Why are these needed for only this model, and not for JMR? Conversely, why is JMR MODFLOW data included in Table 2, but not MSJ-LC MODFLOW data?
Line 181: What is meant by a ‘geographic connection’? Was the data reprojected using GIS?
Line 187: How were the averages weighted? Based on layer thickness, or something else?
Figure 2: Why are we seeing geologic units for one aquifer, but hydraulic conductivity for the other? I'm wondering if the intention is to show two different types of setup/ simulation depending on your aims, but if so, that has not been made clear in the text.
Figure 3: This figure is very blurry.
Line 233-235 and Figure 4: I am confused by the differentiation between options A and B. A makes sense since water will percolate downwards from the unsaturated soil profile; in B, why is exchange only possible from aquifer to soil profile? This implies that the water table can only rise in this case?
Lines 252-261 and Figure 5: On what scale are the HRUs constructed? Are they vectors? Is each agricultural field its own HRU? What else defines an HRU?
Additionally, I thought that the cell size was 500 m (from line 162), but the quoted HRU sizes in this paragraph don’t match with that.
Figure 8: This figure has helpful information, but I feel like I, as the reader, am having to do too much work. Having a reference for what all of these abbreviations mean closer to the image e.g. in the caption would help a lot.
Lines 403-406: What is PEST++? Is it different from PEST? While references are provided, a very brief explanation would be helpful.
Figure 10: What is recharge in all of the areas that are left white?
Figure 12: Is this the change from the model or from the well data? How do the two compare to one another?

Citation: https://doi.org/10.5194/egusphere-2025-300-RC2
- AC2: 'Reply on RC2', Ryan Bailey, 02 Jun 2025
  
  We have replied to all comments from RC2 in the attached document.
  Black text = reviewer text
  Red text = author response
  Blue text = text from the manuscript, used in author response.
  
  Citation: https://doi.org/10.5194/egusphere-2025-300-AC2

Interactive discussion

Status: closed

CC1: 'Comment on egusphere-2025-300', Nima Zafarmomen, 02 Apr 2025

This manuscript presents a technically robust and well-structured integration of SWAT+ with MODFLOW-NWT, offering a significant modeling advancement for coupled surface-subsurface hydrology in managed watersheds.
The authors clearly state that the purpose is to introduce the model, not to fully calibrate it. However, for a model intended to support decision-making in complex, managed basins, stronger evidence of predictive accuracy (e.g., calibrated groundwater heads, fluxes, or uncertainty estimates) would reinforce its credibility.
NSE values are not reported for groundwater head simulations. Including these would help assess model robustness.
The use of a transfer function to represent vadose zone processes is a limitation. The authors should elaborate more on how this simplification affects recharge timing and suggest pathways for integrating UZF in future versions.

While MODFLOW can simulate confined aquifers, the manuscript uses only unconfined systems. Given the relevance of confined systems in agricultural basins, this limits the model's broader applicability and should be more explicitly acknowledged.
The poor performance (NSE = –0.31) in upstream segments of the San Joaquin River is concerning. Although this is not the paper’s central focus, the authors should discuss potential reasons (e.g., inflow data, canal routing) and whether additional calibration steps might help.
While the model runs are reasonably fast, the authors could include more benchmarks or comparisons with SWAT+ alone to quantify the cost-benefit of integrating MODFLOW, especially for large-scale applications.

Citation: https://doi.org/10.5194/egusphere-2025-300-CC1
RC1:
'Comment on egusphere-2025-300', Gerrit H. de Rooij, 07 Apr 2025

Dear editor, dear authors,
I reviewed the paper in detail, and the tutorial cursorily. The case for the model introduced by the authors is made convincingly, and the model clearly falls within the scope of GMD.

Although the model consists of existing components that have been augmented and linked, there is enough new material in the paper to justify publication in GMD. The relevance for practical applications is explained well by the authors.

The tutorial could use a few more references (including links) to established methods/techniques/software that the authors used. Examples of established software used are Notepad++ and QGIS, which are given without link or reference. Several links to other useful resources are given in the text of the tutorial, but not included in the reference list, which is a bit inconvenient.

The tutorial explains the working principle of PEST very well, which was quite helpful. But the Morris method used for the global sensitivity analysis used by PEST++ appears without explanation or reference.

All in all, I have no major comments. A few unclarities and inconsistencies appear in the text, highlighted in the minor comments directly inserted into the text. I suspect the inconsistencies are textual, rather than substantial, and can therefore be removed by a suitable rewriting of the text. I do not think the model code itself needs corrections. If this assessment proves to be correct, minor revisions should be sufficient.

Sincerely,
Gerrit de Rooij

Citation: https://doi.org/10.5194/egusphere-2025-300-RC1
- AC1: 'Reply on RC1', Ryan Bailey, 02 Jun 2025
  
  We have replied to all comments from RC1 in the attached document.
  Black text = reviewer text
  Red text = author response
  Blue text = text from the manuscript, used in author response.
  
  Citation: https://doi.org/10.5194/egusphere-2025-300-AC1
RC2:
'Comment on egusphere-2025-300', Kerry Callaghan, 22 Apr 2025

This paper presents a new coupling of two existing models, SWAT+ (watershed hydrology) and MODFLOW-NWT (groundwater flow). The coupled model is more hydrologically complete than either of the models on their own, and would be a useful tool for many regional hydrological studies. The paper falls within the scope of GMD, and I believe that it will be fit for publication after minor revisions have been completed. I think that this paper and the tool it describes are useful, but certain explanations need to be clarified and figures in particular need some work to aid the reader’s understanding.
This paper provides an updated coupling after a 2020 paper (referenced Bailey et al., 2020a in the paper) already linked SWAT+ and MODFLOW. The motivation given for the new paper is that SWAT+ has been modified extensively over the past 5 years. One question that I have is whether the new linkage will be able to absorb future improvements to SWAT+, or will brand new couplings be required every few years as modifications continue?
The paper links to a tutorial document, executable file, and sample input data to attempt reconstructions of the simulations presented. I appreciate this effort and the inclusion of the sample data. However, as a Linux user, I was unable to run the .exe file. Source code is provided; it would be extremely helpful to include compilation instructions.
Line 90: The paragraph starting at this line seems repetitive and unnecessary. Can it be integrated into the above paragraph?
Figure 1: This This figure needs some work, e.g. both the Arkansas River and the San Joaquin River have the same symbology, so maybe it would work better to use the same symbol and then just specify on which panel does it refer to which river. What are the small black dots on panel B? Why does panel C have the additional aquifer information while panel A does not?
Line 163 states that for the JMR, the aquifer boundary is the same as the watershed boundary (Fig 1B), but Fig 1B does not actually show this – it would be clearer if explicitly shown or mentioned in the caption.
Are aquifer boundaries shown in this figure and mentioned in the text from an external dataset? Table 2 includes aquifer thickness but it is not clear if this dataset also defines the boundaries.
Line 173 begins a description of population grid cell properties for the MSJ-LC model. Why are these needed for only this model, and not for JMR? Conversely, why is JMR MODFLOW data included in Table 2, but not MSJ-LC MODFLOW data?
Line 181: What is meant by a ‘geographic connection’? Was the data reprojected using GIS?
Line 187: How were the averages weighted? Based on layer thickness, or something else?
Figure 2: Why are we seeing geologic units for one aquifer, but hydraulic conductivity for the other? I'm wondering if the intention is to show two different types of setup/ simulation depending on your aims, but if so, that has not been made clear in the text.
Figure 3: This figure is very blurry.
Line 233-235 and Figure 4: I am confused by the differentiation between options A and B. A makes sense since water will percolate downwards from the unsaturated soil profile; in B, why is exchange only possible from aquifer to soil profile? This implies that the water table can only rise in this case?
Lines 252-261 and Figure 5: On what scale are the HRUs constructed? Are they vectors? Is each agricultural field its own HRU? What else defines an HRU?
Additionally, I thought that the cell size was 500 m (from line 162), but the quoted HRU sizes in this paragraph don’t match with that.
Figure 8: This figure has helpful information, but I feel like I, as the reader, am having to do too much work. Having a reference for what all of these abbreviations mean closer to the image e.g. in the caption would help a lot.
Lines 403-406: What is PEST++? Is it different from PEST? While references are provided, a very brief explanation would be helpful.
Figure 10: What is recharge in all of the areas that are left white?
Figure 12: Is this the change from the model or from the well data? How do the two compare to one another?

Citation: https://doi.org/10.5194/egusphere-2025-300-RC2
- AC2: 'Reply on RC2', Ryan Bailey, 02 Jun 2025
  
  We have replied to all comments from RC2 in the attached document.
  Black text = reviewer text
  Red text = author response
  Blue text = text from the manuscript, used in author response.
  
  Citation: https://doi.org/10.5194/egusphere-2025-300-AC2

Peer review completion

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

AR by Ryan Bailey on behalf of the Authors (02 Jun 2025) Author's response Author's tracked changes Manuscript

ED: Publish as is (04 Jun 2025) by Charles Onyutha

AR by Ryan Bailey on behalf of the Authors (28 Jul 2025) Author's response Manuscript

Journal article(s) based on this preprint

05 Sep 2025

SWAT+MODFLOW: a new hydrologic model for simulating surface-subsurface flow in managed watersheds

Ryan T. Bailey, Salam Abbas, Jeffrey G. Arnold, and Michael J. White

Geosci. Model Dev., 18, 5681–5697, https://doi.org/10.5194/gmd-18-5681-2025,https://doi.org/10.5194/gmd-18-5681-2025, 2025

Short summary

Ryan Bailey, Salam Abbas, Jeffrey Arnold, and Michael White

Supplement

https://doi.org/10.5194/egusphere-2025-300-supplement

Data sets

SWAT+MODFLOW model: code and example models Ryan Bailey https://zenodo.org/records/14674981

Model code and software

SWAT+MODFLOW model: code and example models Ryan Bailey https://zenodo.org/records/14674981

Ryan Bailey, Salam Abbas, Jeffrey Arnold, and Michael White

Viewed

Total article views: 1,339 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
991	330	18	1,339	129	16	30

HTML: 991
PDF: 330
XML: 18
Total: 1,339
Supplement: 129
BibTeX: 16
EndNote: 30

Views and downloads (calculated since 20 Mar 2025)

Month	HTML	PDF	XML	Total
Mar 2025	94	33	3	130
Apr 2025	122	76	4	202
May 2025	72	55	1	128
Jun 2025	137	71	7	215
Jul 2025	103	51	1	155
Aug 2025	345	37	2	384
Sep 2025	118	7	0	125

Cumulative views and downloads (calculated since 20 Mar 2025)

Month	HTML	PDF	XML	Total
Mar 2025	94	33	3	130
Apr 2025	122	76	4	202
May 2025	72	55	1	128
Jun 2025	137	71	7	215
Jul 2025	103	51	1	155
Aug 2025	345	37	2	384
Sep 2025	118	7	0	125

Viewed (geographical distribution)

Total article views: 1,345 (including HTML, PDF, and XML) Thereof 1,345 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 05 Sep 2025

Download

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

Preprint (3025 KB)
Metadata XML

Short summary

Water managers often make use of computer models to assess a region’s water supply under future conditions and management scenarios. This article introduces a new computer model that combines a land surface model (SWAT+) and a groundwater model (MODFLOW) and shows how it can be applied to managed, irrigated watersheds. This new model can be used for regions that rely on both surface water and groundwater for drinking water, agriculture, and industry.


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