HESS Opinions: Applied hydrologic models in the era of machine learning &ndash; retain, revamp, reconcile, or replace?

Williams, Delanie; Kumar, Mukesh; van Werkhoven, Katie; Clark, Martyn; Wilson, Christopher; Miller, Paul

doi:10.5194/egusphere-2026-583

Preprints

https://doi.org/10.5194/egusphere-2026-583

Preprints

19 Feb 2026

| 19 Feb 2026

HESS Opinions: Applied hydrologic models in the era of machine learning – retain, revamp, reconcile, or replace?

Delanie Williams, Mukesh Kumar, Katie van Werkhoven, Martyn Clark, Christopher Wilson, and Paul Miller

Abstract. Despite advancements in the performance of machine learning (ML) based hydrologic models, some institutions are hesitant to pursue ML as a replacement for existing conceptual or process-based hydrologic models in many applications. In several of these circumstances, traditional hydrologic models continue to be favored due to their familiarity, reliability, interpretability, established performance benchmarks under varied settings, availability of detailed training modules and a trained workforce, as well as close integration with data, processing, and decision-making pipelines. Recognizing these advantages, this perspective argues for two pragmatic and institutionally compatible paths forward for integration of ML within applied models: (1) reconciling ML as a complementary option in applied hydrologic modeling workflows; and (2) revamping or upskilling hydrologic modeling workflows using ML. To support this perspective, we highlight key opportunities where ML can be used as a tool to enhance results across various stages of the model implementation and operational workflow including data pre-processing, parameter calibration, parameter transferability, data assimilation, solver enhancement, accelerating scenario simulations and post-processing. Each of these two integration strategies can be implemented into current applied model frameworks, thereby combining the strengths of both physical modeling and ML. These strategies can help overcome current bottlenecks and address institutional needs of continuity and compatibility, while also offering the potential to improve model performance with ML.

Received: 30 Jan 2026 – Discussion started: 19 Feb 2026

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

Delanie Williams, Mukesh Kumar, Katie van Werkhoven, Martyn Clark, Christopher Wilson, and Paul Miller

Status: final response (author comments only)

CC1:
'Comment on egusphere-2026-583', Nima Zafarmomen, 25 Feb 2026

This manuscript presents a timely and well-articulated opinion on the evolving role of applied hydrologic models in the era of ML. The paper is well structured, written in a clear and accessible style, and supported by illustrative examples spanning forecasting, planning, and decision-support contexts. The proposed roadmap for ML integration is particularly valuable, as it frames adoption not only as a technical evolution but also as an institutional and cultural transition. Overall, the manuscript makes a meaningful contribution to an important and ongoing discussion. However, several issues should be addressed to further strengthen the clarity, rigor, and practical impact of the paper.
Comments
1- The manuscript would benefit from a more explicit decision framework clarifying when users should retain, revamp, reconcile, or replace existing modeling approaches. While the conceptual distinctions are helpful, readers will seek clearer decision criteria or guiding principles. Incorporating a structured comparison (e.g., a decision matrix based on data availability, interpretability needs, regulatory constraints, computational cost, and operational risk) would substantially improve the manuscript’s applicability.
2- The paper surveys a wide range of models, tasks, and ML integration opportunities. Although informative, the breadth risks diluting the central message. The manuscript would be strengthened by prioritizing or highlighting the most impactful and realistic integration pathways (e.g., calibration acceleration, surrogate modeling, bias correction, forcing-data improvement). This would enhance focus and provide clearer guidance for applied users.
3- Sections discussing LLMs)introduce interesting perspectives but would benefit from clearer boundaries regarding current capabilities versus future potential. Framing LLM-related discussions explicitly as emerging prospects would improve precision and avoid overgeneralization. I do recommend to consider papers such as “Can large language models effectively reason about adverse weather conditions?”, which reflects an active and relevant research frontier. Additionally, other emerging computational paradigms could be briefly acknowledged to broaden the forward-looking perspective. For example, quantum computing is increasingly discussed in environmental modeling contexts. The authors may consider citing recent developments such as "HydroQuantum: A new quantum-driven Python package for hydrological simulation" as an example of exploratory directions that, while still nascent, may influence future modeling workflows.

Citation: https://doi.org/10.5194/egusphere-2026-583-CC1
- AC3: 'Response to Reviewer 1 (Nima Zafarmomen) Comments', Delanie Williams, 12 Jul 2026
  
  Thank you for taking the time to leave suggestions and comments on our opinion article. Below we have addressed all original comments (bold), and answered with responses in standard text and updated manuscript quotes in italics.
  Summary: This manuscript presents a timely and well-articulated opinion on the evolving role of applied hydrologic models in the era of ML. The paper is well structured, written in a clear and accessible style, and supported by illustrative examples spanning forecasting, planning, and decision-support contexts. The proposed roadmap for ML integration is particularly valuable, as it frames adoption not only as a technical evolution but also as an institutional and cultural transition. Overall, the manuscript makes a meaningful contribution to an important and ongoing discussion. However, several issues should be addressed to further strengthen the clarity, rigor, and practical impact of the paper.
  Response: Thank you for recognizing the value and timeliness of this perspective and for providing thoughtful, detailed feedback. We have carefully addressed each of the concerns raised, and we believe that the revisions have substantially strengthened the manuscript.
  Major Comments
  Comment 1. The manuscript would benefit from a more explicit decision framework clarifying when users should retain, revamp, reconcile, or replace existing modeling approaches. While the conceptual distinctions are helpful, readers will seek clearer decision criteria or guiding principles. Incorporating a structured comparison (e.g., a decision matrix based on data availability, interpretability needs, regulatory constraints, computational cost, and operational risk) would substantially improve the manuscript’s applicability.
  Response: Thank you for this insightful observation about improving the usability of the decision framework in operational contexts. We agree that the “roadmap” should incorporate more explicit decision-making trade-offs to further guide users in model choice and development. Within section 4, we have further defined each technique (i.e. retention, replacement, reconciliation, and revamping) and described in what cases they are most aligned based off institutional resources. Furthermore, in subsections 4.1 and 4.2, we inserted specific questions to guide readers in selecting the appropriate technique for their organization. These questions are as follows (lines 273-278):
  “Key questions to consider include: (i) does the existing model adequately meet operational needs? (ii) Which limitations, when addressed, would improve the model’s utility and/or effectiveness? (iii) Are regulatory approvals, established workflows, or stakeholder expectations tied to the existing modeling framework? (iv) Does prior research demonstrate substantial improvements through ML for some or all of the model’s intended applications? (v) Are sufficient financial, computational, and personnel resources available to support the partial or complete replacement of existing model workflows? Finally, (iv) would the intended implementation technique be sustainable over time? Comment 2. The paper surveys a wide range of models, tasks, and ML integration opportunities. Although informative, the breadth risks diluting the central message. The manuscript would be strengthened by prioritizing or highlighting the most impactful and realistic integration pathways (e.g., calibration acceleration, surrogate modeling, bias correction, forcing-data improvement). This would enhance focus and provide clearer guidance for applied users.”
  Comment 2. The paper surveys a wide range of models, tasks, and ML integration opportunities. Although informative, the breadth risks diluting the central message. The manuscript would be strengthened by prioritizing or highlighting the most impactful and realistic integration pathways (e.g., calibration acceleration, surrogate modeling, bias correction, forcing-data improvement). This would enhance focus and provide clearer guidance for applied users.
  Response: Thank you for this comment. The intention of the wide range of models and ML integration opportunities mentioned is twofold: 1) We wish to recognize the wide variety of models which are used within operations and the variety of integration objectives which could be desired by the community. 2) We want to highlight that there are numerous avenues for integrating ML within modeling workflows. While we have sharpened the focus of the paper through our revisions, we have intentionally retained the descriptions of the models and tasks to provide a comprehensive overview and enhance the paper's relevance and applicability across the field.
  Comment 3. Sections discussing LLMs introduce interesting perspectives but would benefit from clearer boundaries regarding current capabilities versus future potential. Framing LLM-related discussions explicitly as emerging prospects would improve precision and avoid overgeneralization. I do recommend to consider papers such as “Can large language models effectively reason about adverse weather conditions?”, which reflects an active and relevant research frontier. Additionally, other emerging computational paradigms could be briefly acknowledged to broaden the forward-looking perspective. For example, quantum computing is increasingly discussed in environmental modeling contexts. The authors may consider citing recent developments such as "HydroQuantum: A new quantum-driven Python package for hydrological simulation" as an example of exploratory directions that, while still nascent, may influence future modeling workflows.
  Response: Thank you for this comment. We have revised the manuscript LLM discussion within section 3.2 to further emphasize existing LLM research within meteorology and hydrology by adding additional sources. In lines 222-229, we expanded as follows:
  “Emergency communication frameworks can be supported with large language models (LLM). Within meteorology, various LLMs have demonstrated an ability to discern adverse weather conditions from National Weather Serve (NWS) text data, suggesting that real-time monitoring and analysis of disaster events is on the horizon (Zafarmomen and Samadi, 2025). LLMs have also shown promise with emergency-related message interpretation and classification, indicating future ML potential in 911 dispatch operation improvement and personalized disaster information inquiries (Otal et al., 2024). ML in decision support can assist in the comprehension of flood warnings; for lay-people, trained flood risk large language models (LLM) can answer flood-related questions and therefore reduce an individual’s flood risk (Zhu et al., 2024).”
  In this way, we have attempted to limit the discussion to existing potential without emphasizing potential avenues, which are expansive and couldn’t be explored in the necessary detail.
  We sincerely thank the reviewer for their time and thoughtful comments. Their constructive feedback has helped us strengthen the manuscript.
  
  Citation: https://doi.org/10.5194/egusphere-2026-583-AC3
RC1:
'Comment on egusphere-2026-583', Anonymous Referee #1, 16 May 2026
This manuscript presents a perspective on integrating machine learning (ML) into applied hydrologic modeling workflows, arguing for “reconciliation” or “revamping” rather than replacement of legacy models. The topic is timely and practically relevant, as many operational agencies are indeed grappling with how to incorporate ML advances. However, I have several concerns regarding the depth of the contribution, the clarity of the intended audience, and whether the proposed roadmap is actionable enough to move the field forward. My overall impression is that the manuscript largely synthesizes well-known challenges and opportunities without putting forward substantially new ideas or demonstrating that the proposed workflow would be effective in practice. The arguments would benefit from more specificity, more critical assessment of limitations, and clearer articulation of what this perspective adds beyond what the community already understands.
Major Comments
Table 1: The purpose of Table 1 is unclear. Many widely used models such as VIC, SUMMA, and WRF-Hydro, which are mentioned in the text, are absent. In its current form, the table reads as an ad hoc selection of models without a clear rationale for inclusion or exclusion. I would suggest organizing models by category (e.g., lumped rainfall-runoff models, distributed watershed models, hydrodynamic models, integrated surface-subsurface models) and clearly stating the selection criteria. If the intent is simply to provide illustrative examples, the authors should state this explicitly and trim the table accordingly.

Section 4.3 Step 3. Select an Appropriate ML Technique: This is proposed as a step in the road map. In practice, this is challenging and often cannot be resolved a priori. The best-performing ML approach varies depending on the use case, data availability, catchment characteristics, and the nature of the modeling deficiency being addressed. There is rarely a single "correct" technique, and the selection process typically requires iterative experimentation and benchmarking. As written, Step 3 risks giving practitioners an unrealistically linear impression of a complex and iterative process.

Section 4: It is unclear who is the intended audience of the proposed road map. The framework is presented at a high level that may be too general for any specific audience (academia, agencies, or industry to find directly actionable. Each of these communities faces different constraints. For example, agencies may lack ML expertise and computational infrastructure; academics may lack access to operational systems and institutional buy-in. The manuscript would be strengthened by explicitly addressing the skills, resources, and institutional conditions required to implement the proposed workflow. I would encourage the authors to ground the framework with more concrete guidance, perhaps including rough resource estimates, example timelines, or case studies where similar integration has been attempted.

As an Opinions paper, this manuscript is expected to offer a clear, forward-looking perspective that advances the community's thinking. While the framing around “retain, revamp, reconcile, or replace” is effective, the individual ideas presented (using ML for data gap-filling, calibration, surrogate modeling, etc.) are already well-established in the literature. I would encourage the authors to more clearly articulate what is new in their perspective.

Minor comments
Table 1 lists “select municipal water treatment plants” as the typical users of MIKE SHE, which could be misleading. MIKE SHE is a fully integrated, physically-based hydrologic model whose user base is considerably broader than municipal water treatment.

Lines 171–178 (Section 3.2, solver enhancement): The body of ML for accelerating ODE/PDE solutions is substantively different from the data-driven hydrologic modeling discussed elsewhere in the paper and falls more naturally under the umbrella of physics-informed methods, and the authors should use this term explicitly. This would help the reader connect this discussion to a rich and rapidly growing literature (e.g., physics-informed neural networks or PINN, neural operators)

General proofreading: The manuscript would benefit from a careful proof reading. For example, missing period after outputs in line 141. The sentence starting line 140 largely repeats the point already made two lines above.
Citation: https://doi.org/10.5194/egusphere-2026-583-RC1
- AC1: 'Response to Reviewer 2 (Anonymous Referee #1) Comments', Delanie Williams, 12 Jul 2026
  
  Summary: This manuscript presents a perspective on integrating machine learning (ML) into applied hydrologic modeling workflows, arguing for “reconciliation” or “revamping” rather than replacement of legacy models. The topic is timely and practically relevant, as many operational agencies are indeed grappling with how to incorporate ML advances. However, I have several concerns regarding the depth of the contribution, the clarity of the intended audience, and whether the proposed roadmap is actionable enough to move the field forward. My overall impression is that the manuscript largely synthesizes well-known challenges and opportunities without putting forward substantially new ideas or demonstrating that the proposed workflow would be effective in practice. The arguments would benefit from more specificity, more critical assessment of limitations, and clearer articulation of what this perspective adds beyond what the community already understands.
  Response: Thank you for recognizing the relevance and timeliness of this perspective and for providing thoughtful, detailed feedback. We have carefully addressed each of the concerns raised, and we believe that the revisions have substantially strengthened the manuscript.
  Major Comments
  
  Comment 1. Table 1: The purpose of Table 1 is unclear. Many widely used models such as VIC, SUMMA, and WRF-Hydro, which are mentioned in the text, are absent. In its current form, the table reads as an ad hoc selection of models without a clear rationale for inclusion or exclusion. I would suggest organizing models by category (e.g., lumped rainfall-runoff models, distributed watershed models, hydrodynamic models, integrated surface-subsurface models) and clearly stating the selection criteria. If the intent is simply to provide illustrative examples, the authors should state this explicitly and trim the table accordingly.
  Response: Thank you for this comment. While the table was originally providing illustrative examples, we acknowledge that the ultimate purpose and structure of the table does not synthesize existing models well, nor contribute meaningfully to the argument. The table has been removed from the manuscript, and in its place, we have listed illustrative examples of models used in applied settings.
  Comment 2. Section 4.3 Step 3. Select an Appropriate ML Technique: This is proposed as a step in the road map. In practice, this is challenging and often cannot be resolved a priori. The best-performing ML approach varies depending on the use case, data availability, catchment characteristics, and the nature of the modeling deficiency being addressed. There is rarely a single "correct" technique, and the selection process typically requires iterative experimentation and benchmarking. As written, Step 3 risks giving practitioners an unrealistically linear impression of a complex and iterative process.
  Response: Thank you for this comment on section 4.3. We concur that Step 3 of the framework came across as too high-level and suggests a simple process. Within the revised manuscript, we have added the following in lines 300-302 to emphasize how the process may require iteration and benchmarking.
  “Importantly, identifying an appropriate ML method, whether for wholesale replacement, reconciliation, or revamp, could require an iterative process rather than a straightforward decision, unless clear guidance exists from prior studies. Organizations may need to evaluate and benchmark multiple candidate methods before selecting the most appropriate solution.”
  Comment 3. Section 4: It is unclear who is the intended audience of the proposed road map. The framework is presented at a high level that may be too general for any specific audience (academia, agencies, or industry to find directly actionable. Each of these communities faces different constraints. For example, agencies may lack ML expertise and computational infrastructure; academics may lack access to operational systems and institutional buy-in. The manuscript would be strengthened by explicitly addressing the skills, resources, and institutional conditions required to implement the proposed workflow. I would encourage the authors to ground the framework with more concrete guidance, perhaps including rough resource estimates, example timelines, or case studies where similar integration has been attempted.
  Response: Thank you for this comment. The intended audience of this manuscript is agencies and industries, so they can apply the research findings within academia in operational settings. However, we acknowledge that this was not clearly communicated. Within section 4, we have further defined the suggested institutional conditions for each ML implementation process. In doing so, we aim to maintain the existing broad audience, while grounding the implementation techniques with clear considerations. From lines 242-260, we have the following excerpt:
  “-- Retention may be most appropriate when existing models adequately meet operational requirements, regulatory constraints, and stakeholder expectations. If the marginal benefits of ML integration do not justify transition costs and risks, maintaining the current modeling framework may be the preferred option.
  -- Replacement would become a viable option when ML models can fully satisfy analytical and operational needs, and when performance gains (e.g., improved accuracy, reduced simulation time, etc.) from it are substantial. This pathway would be most appropriate when computational infrastructure and in-house expertise exist to support a transition to an ML-centric system. It would be particularly suitable in settings where intermediate process states are not essential for decision-making, regulatory barriers are minimal or manageable, and interpretability requirements are less stringent.
  -- Reconciliation offers a transitional pathway in which ML-based models are run in parallel with existing process-based systems. Beyond methodological benefits like ensemble outputs or faster simulation results, this pathway may also allow organizations to compare performance, evaluate strengths and weaknesses, and build organizational confidence in ML approaches over time. Insights gained from parallel operation can inform gradual integration or eventual replacement. However, this approach would likely require more resource needs than that for the Replacement pathway, due to its need to sustain dual-track systems and associated operational costs.
  -- Revamping may be the most practical pathway for many organizations. In this approach, selected components of existing modeling workflows are enhanced or replaced with ML-based modules, enabling a continuum of modernization rather than a binary shift between applied modeling and ML. This pathway is expected to be less resource-intensive than full replacement and could be undertaken in piecemeal manner consistent with the availability of resources and institutional support at a given time. It is to be noted that it may still require investment in staff training and incremental system upgrades.”
  Comment 4. As an Opinions paper, this manuscript is expected to offer a clear, forward-looking perspective that advances the community's thinking. While the framing around “retain, revamp, reconcile, or replace” is effective, the individual ideas presented (using ML for data gap-filling, calibration, surrogate modeling, etc.) are already well-established in the literature. I would encourage the authors to more clearly articulate what is new in their perspective.
  Response: Thank you for this advice. To more clearly articulate the novelty of our perspective, we have revised section 1 to more clearly articulate the purpose of our opinion article. We have reworded and added to lines 67-76.
  “The goal of this perspective is to highlight the key challenges that currently hinder the integration of ML within some current operational settings, despite their researched promise of improved predictive accuracy. The focus is not to discuss the adequacy of ML methods relative to conceptual or physics-based models, as such comparisons have been extensively explored in the literature (Painter and Destouni, 2026; Yan et al., 2026; Kok Poh et al., 2022; Xu and Liang, 2021a; Sun and Scanlon, 2019). Rather, we frame this perspective around how organizations that depend on operational hydrologic models should navigate modernization decisions in the era of rapidly advancing ML capabilities, particularly when complete replacement by ML is impractical, undesirable, or infeasible. We specifically highlight the need for organizations to systematically evaluate opportunities for retaining, revamping, reconciling, or replacing existing modelling systems with ML-based approaches. In doing so, we shift the focus from model competition toward defining practical ML pathways for integration within applied model workflows.”
  Minor comments
  
  Comment 1. Table 1 lists “select municipal water treatment plants” as the typical users of MIKE SHE, which could be misleading. MIKE SHE is a fully integrated, physically-based hydrologic model whose user base is considerably broader than municipal water treatment.
  Response: Thank you for pointing this out. MIKE-SHE can be used outside of what is labeled within the table. To remedy this, we have removed the table which described too few of the capabilities of MIKE-SHE and instead have replaced it with a sentence simply describing that it is used in applied modeling workflows.
  Comment 2. Lines 171–178 (Section 3.2, solver enhancement): The body of ML for accelerating ODE/PDE solutions is substantively different from the data-driven hydrologic modeling discussed elsewhere in the paper and falls more naturally under the umbrella of physics-informed methods, and the authors should use this term explicitly. This would help the reader connect this discussion to a rich and rapidly growing literature (e.g., physics-informed neural networks or PINN, neural operators)
  Response: Thank you for this suggestion. We have added a line within section 3.2 which incorporates the language “physics-informed machine learning” to help the reader connect this specific integration method with the wider literature.
  Comment 3. General proofreading: The manuscript would benefit from a careful proof reading. For example, missing period after outputs in line 141. The sentence starting line 140 largely repeats the point already made two lines above.
  Response: Thank you for this comment. Both the missing period in line 141 and the sentence in line 140 have been adjusted accordingly. The manuscript has since been carefully proofread to ensure professional quality.
  Thank you again for your time and thoughtful comments. Your feedback has significantly helped us improve the manuscript.
  
  Citation: https://doi.org/10.5194/egusphere-2026-583-AC1
RC2:
'Comment on egusphere-2026-583', Anonymous Referee #2, 19 May 2026

General comments
The use of AI/ML in hydrology has substantially grown in the past decade in academic contexts (i.e., numerous publications), but its adoption for operational purposes remains relatively slow. This paper raises an important issue of why there is hesitation in adopting AI in applied models, with a catchy title that poses interesting questions: to retain, revamp, reconcile, or replace? The authors offer good insights into why legacy models and workflows persist despite the superior performance demonstrated by recent ML models.
However, in the discussion the paper does not substantively address all options proposed (retain, revamp, reconcile, or replace). The workflow presented (identify limitations, apply an approach, test, and gradually incorporate) is simplistic. It appears incremental, and is essentially retaining and revamping long-established hydrologic model improvement workflows, where traditional model structure or parameterization improvements are instead replaced with ML integration or architecture choice. Overall, the paper presents a narrow view of incorporating ML into existing hydrologic modeling workflows. If we look at the evolution of weather forecasting, ML methods are now rapidly being adopted as the primary forecasters, and process model outputs are being used as data for training. The same shift could occur soon in hydrology, potentially rendering current models and workflows less useful for large-scale, high-resolution, ML-based operational predictions except in generation of synthetic training data. This and other “replace”/reconcile possibilities are not discussed.
Hence, the current discussion seems superficial, and reads more like a workshop summary. The authors also overlook recent advances such as agentic AI, coding assistants, generative AI, and foundation models. Large Language Models (LLMs) are mentioned in passing, mostly from a perspective of improving model outputs. It also neglects significant progress in hybrid process-ML architectures, such as differentiable modeling that can improve the National Water Model (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024WR038928) and USGS efforts to implement process-guided deep learning into their national models. The conclusions advocate for the “retain” workflows, without making a strong case for why that is better than some of the more advanced ML methods that have shown demonstrable improvements in prediction accuracy, resolution and speed over traditional methods.
This paper would be more interesting, if it presented a more in-depth analysis of the barriers faced by hydrological modelers in integrating AI. While workflows and upskilling/training are mentioned, the discussion does not touch on other issues such as lack of data, limited GPU compute resources, and the difficulty of keeping up with rapidly evolving AI developments and tools. In particular, the authors do not discuss data-related challenges, including the lack of benchmark datasets, and time-consuming aspects such as data wrangling, quality control, and movement of large data to compute nodes. Most ML demonstrations in hydrology rely on a small set of benchmarks (e.g., CAMELS and its derivatives), curated monitoring network data (e.g., from USGS) or remote sensing products. For operational models, the authors could expand on their brief mentions of interpretability and trustworthiness. Specifically, it would be valuable to discuss validation methods and thresholds for accuracy (or other relevant metrics) required to make the decision about which modeling pathway is the best, and how to gain the trust of operational practitioners. In general, hydrology lacks sufficient benchmarks for model intercomparisons, and hence it is currently challenging to determine the failure modes of ML (or process) models. It is also unclear what timelines are being suggested for slow ML adoption (months, years?), and this suggestion fails to recognize how quickly AI methods are improving and being adopted for many other purposes.
Overall, this paper adds very little in terms of new suggestions for moving the field forward beyond what has been covered previously. There are already numerous opinions, reviews, and reports on the application of ML in hydrologic science and modeling, and the paths forward with integration of process-based and ML models. Below is a list of examples of such papers going back several years. None (other than Nearing et al. (2020) of these are cited in this manuscript.
Painter and Destouni (2026) https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2026WR043509
Yan et al. (2026): https://www.sciencedirect.com/science/article/pii/S1674237026000025
Zhi et al. (2024) https://www.nature.com/articles/s44221-024-00202-z
Varadharajan et al. (2022): https://onlinelibrary.wiley.com/doi/abs/10.1002/hyp.14565
Xu and Liang (2021) https://wires.onlinelibrary.wiley.com/doi/abs/10.1002/wat2.1533
Nearing et al. (2020) https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2020WR028091
Fleming and Gupta (2020) https://physicstoday.aip.org/features/the-physics-of-river-prediction
Sun and Scanlon (2019) https://iopscience.iop.org/article/10.1088/1748-9326/ab1b7d
Specific comments
Table 1 is not useful or comprehensive as a list of process models. There is a huge number of process models used in hydrology (see the extensive list of hydrologic models maintained by the CSDMS https://csdms.colorado.edu/wiki/Hydrological_Models). Alternatively, Figure 1 in Yan et al. (2026) provides a useful overview of the trajectory of hydrologic model development. The main point to make is that there is a huge diversity of process models at varying levels of complexity, and the opportunities for integration with AI can vary a lot depending on the model type. This paper could explore that concept further.
The workflow sections are very short, lacking depth, and should be expanded substantially, as the primary contribution of the paper. The integration objectives and criteria for choosing an ML approach are big topics that can be elaborated on further.
The paper lacks sufficient referencing for an opinion piece. For example, it includes only three references to single-variable machine learning (ML) models, despite the existence of hundreds of publications demonstrating ML model skill across a diverse range of variables. As previously mentioned, numerous review papers that cite hundreds of these studies are available and should be included.

Citation: https://doi.org/10.5194/egusphere-2026-583-RC2
- AC2:
  'Response to Reviewer #3 (Anonymous Referee #2 Comments', Delanie Williams, 12 Jul 2026
  Thank you for taking the time to leave thoughtful comments and suggestions. Within our response, reviewer comments are bolded, responses are standard text, and manuscript quotes are italicized.
  The use of AI/ML in hydrology has substantially grown in the past decade in academic contexts (i.e., numerous publications), but its adoption for operational purposes remains relatively slow. This paper raises an important issue of why there is hesitation in adopting AI in applied models, with a catchy title that poses interesting questions: to retain, revamp, reconcile, or replace? The authors offer good insights into why legacy models and workflows persist despite the superior performance demonstrated by recent ML models.
  Response: Thank you for recognizing the relevance of this perspective and for providing thoughtful, detailed feedback. We have carefully addressed each of the concerns raised, and we believe that the revisions have substantially strengthened the manuscript.
  However, in the discussion the paper does not substantively address all options proposed (retain, revamp, reconcile, or replace). The workflow presented (identify limitations, apply an approach, test, and gradually incorporate) is simplistic. It appears incremental, and is essentially retaining and revamping long-established hydrologic model improvement workflows, where traditional model structure or parameterization improvements are instead replaced with ML integration or architecture choice. Overall, the paper presents a narrow view of incorporating ML into existing hydrologic modeling workflows. If we look at the evolution of weather forecasting, ML methods are now rapidly being adopted as the primary forecasters, and process model outputs are being used as data for training. The same shift could occur soon in hydrology, potentially rendering current models and workflows less useful for large-scale, high-resolution, ML-based operational predictions except in generation of synthetic training data. This and other “replace”/reconcile possibilities are not discussed.
  Hence, the current discussion seems superficial, and reads more like a workshop summary. The authors also overlook recent advances such as agentic AI, coding assistants, generative AI, and foundation models. Large Language Models (LLMs) are mentioned in passing, mostly from a perspective of improving model outputs. It also neglects significant progress in hybrid process-ML architectures, such as differentiable modeling that can improve the National Water Model (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024WR038928) and USGS efforts to implement process-guided deep learning into their national models. The conclusions advocate for the “retain” workflows, without making a strong case for why that is better than some of the more advanced ML methods that have shown demonstrable improvements in prediction accuracy, resolution and speed over traditional methods.
  Response: We thank the reviewer for this perceptive comment. In response, we have revised to better articulate the central message of the paper. Specifically, we clarify that revamping and reconciliation are not inherently superior pathways for modernizing applied modeling workflows compared with retaining existing models or replacing them entirely with ML. Rather, we present them as valuable alternatives in situations where existing modeling workflows remain preferable for scientific, operational, or institutional reasons. This clarification is reflected in the revised Abstract, where we now state:
  “In several [….] circumstances, traditional hydrologic models continue to be favored due to their familiarity, reliability, interpretability, established performance benchmarks under varied settings, availability of detailed training modules and a trained workforce, as well as close integration with data, processing, and decision-making pipelines. Recognizing these advantages, this perspective argues for moving beyond a binary framing of either retaining or replacing existing modeling systems. Instead, we emphasize the need to consider additional pragmatic and institutionally compatible modernization pathways in the era of ML. These include: (1) reconciling ML as a complementary option in applied hydrologic modeling workflows; and (2) revamping or upskilling hydrologic modeling workflows using ML. We further argue that the appropriate pathway is not determined solely by predictive accuracy, but also by broader institutional and operational considerations, including regulatory constraints, workforce expertise and readiness, computational and data infrastructure, interpretability requirements, reliance on intermediate process states in decision-making, and tolerance for operational and scientific risk. Building on these considerations, we then present a structured, decision-oriented framework to guide organizations in evaluating and selecting ML integration pathways aligned with their objectives, resources, and institutional constraints.”
  In other words, this perspective does not argue that applied hydrologic models cannot or should not eventually be replaced by ML. Rather, we emphasize that complete replacement is only one of several viable modernization pathways for applied hydrologic modeling workflows in the era of ML. The appropriate pathway should be guided not only by predictive performance but also by organizational constraints, workforce readiness, regulatory requirements, computational and data infrastructure, implementation considerations, and the specific objectives of the modeling application. These are explicitly discussed in Section 4:
  “To evaluate how to modernize applied hydrologic model workflows, organizations must consider not only technical feasibility but also their objectives, available resources, and institutional constraints. Successful integration of ML depends on access to computational infrastructure, an educated workforce, and sustained institutional support. Unlike large technology firms, many water agencies, utilities, consulting organizations, and local governments operate under constrained and inflexible budgets and may not have dedicated GPU resources or specialized ML personnel. As a result, resource availability should be carefully assessed before selecting any integration strategy.”
  Furthermore, we present a roadmap for integration of ML based on the selected pathway. The framework is intended to serve as a practical guide for organizations that rely on applied hydrologic modelling systems as they evaluate and implement ML within their modeling workflows. To this end, we have expanded the descriptions of each implementation pathway and explicitly outlined the considerations and potential barriers that may favor one pathway over another, thereby providing a more balanced and impartial framework for agencies and organizations. An excerpt can be seen from section 4, lines 242-260.
  “-- Retention may be most appropriate when existing models adequately meet operational requirements, regulatory constraints, and stakeholder expectations. If the marginal benefits of ML integration do not justify transition costs and risks, maintaining the current modeling framework may be the preferred option.
  -- Replacement would become a viable option when ML models can fully satisfy analytical and operational needs, and when performance gains (e.g., improved accuracy, reduced simulation time, etc.) from it are substantial. This pathway would be most appropriate when computational infrastructure and in-house expertise exist to support a transition to an ML-centric system. It would be particularly suitable in settings where intermediate process states are not essential for decision-making, regulatory barriers are minimal or manageable, and interpretability requirements are less stringent.
  -- Reconciliation offers a transitional pathway in which ML-based models are run in parallel with existing process-based systems. Beyond methodological benefits like ensemble outputs or faster simulation results, this pathway may also allow organizations to compare performance, evaluate strengths and weaknesses, and build organizational confidence in ML approaches over time. Insights gained from parallel operation can inform gradual integration or eventual replacement. However, this approach would likely require more resource needs than that for the Replacement pathway, due to its need to sustain dual-track systems and associated operational costs.
  -- Revamping may be the most practical pathway for many organizations. In this approach, selected components of existing modeling workflows are enhanced or replaced with ML-based modules, enabling a continuum of modernization rather than a binary shift between applied modeling and ML. This pathway is expected to be less resource-intensive than full replacement and could be undertaken in piecemeal manner consistent with the availability of resources and institutional support at a given time. It is to be noted that it may still require investment in staff training and incremental system upgrades..”
  This paper would be more interesting, if it presented a more in-depth analysis of the barriers faced by hydrological modelers in integrating AI. While workflows and upskilling/training are mentioned, the discussion does not touch on other issues such as lack of data, limited GPU compute resources, and the difficulty of keeping up with rapidly evolving AI developments and tools. In particular, the authors do not discuss data-related challenges, including the lack of benchmark datasets, and time-consuming aspects such as data wrangling, quality control, and movement of large data to compute nodes. Most ML demonstrations in hydrology rely on a small set of benchmarks (e.g., CAMELS and its derivatives), curated monitoring network data (e.g., from USGS) or remote sensing products. For operational models, the authors could expand on their brief mentions of interpretability and trustworthiness. Specifically, it would be valuable to discuss validation methods and thresholds for accuracy (or other relevant metrics) required to make the decision about which modeling pathway is the best, and how to gain the trust of operational practitioners. In general, hydrology lacks sufficient benchmarks for model intercomparisons, and hence it is currently challenging to determine the failure modes of ML (or process) models. It is also unclear what timelines are being suggested for slow ML adoption (months, years?), and this suggestion fails to recognize how quickly AI methods are improving and being adopted for many other purposes.
  Response: We thank the reviewer for this important comment. In response, we have clarified and expanded on the previous research discussing the barriers to ML implementation in operational and applied hydrology. The following paragraphs demonstrate how we expanded on ML implementation barriers, such as multi-variable prediction dynamics, complex decision-making processes, economic factors, computational infrastructure, and data requirements (lines 83-105). The following paragraphs are located in section 2, preceding our discussion of the previously discussed institutional factors (regulations, familiarity, and interpretability).
  “Applied models are used to support a diversity of decisions, involving multiple interacting processes and variables, making it difficult to homogeneously identify ML implementation scenarios. Recent research has demonstrated the efficacy of ML for single hydrologic variable estimation (Ghimire et al., 2021; Malekzadeh et al., 2019; Kratzert et al., 2019) and a few instances of ML models being used in operations (Fleming et al., 2021; Fleming and Goodbody, 2019; Bearup et al., 2024). However, applied model workflows often represent multi-variable dynamics at varying temporal and spatial time scales to facilitate decision making. For example, groundwater management often requires spatially distributed estimates of hydraulic heads, recharge rates, and groundwater-surface water interactions. Similarly, flood risk assessments depend not only on streamflow predictions, which ML has gotten good at (Song et al., 2025), but also on water depths, velocities, inundation extents, and infrastructure impacts. Many applications further involve evaluating future climate, land-use, infrastructure, or management scenarios for which observational analogues could be limited or unavailable. Additional challenges arise from pervasive human modifications to hydrologic systems, including irrigation withdrawals, urban drainage networks, water transfers, and evolving land-use practices. Because these influences are often poorly observed and rapidly changing, developing transferable purely data-driven models remains difficult in many settings. While these considerations do not diminish the potential of ML, they explain why some organizations continue to rely on established hydrologic modeling frameworks while exploring pathways for incremental ML integration.
  The hesitation within organizations to replace complex modeling workflows is further exacerbated by a myriad of institutional factors, such as the economic costs associated with development, workforce training, maintenance, and long-term support of ML-based systems. Replacement of applied models by ML is expected to require continued investments in computing infrastructure, data management, software maintenance, model retraining, and specialized personnel. Given the rapid pace of change in ML methods and technologies, these long-term costs may not be offset by anticipated benefits. This is especially relevant for agencies, utilities, and consulting organizations which operate with smaller budgets. Consequently, organizations with limited resources and low tolerance for operational risk may favour incremental integration of ML within existing workflows rather than wholesale replacement of established modeling systems.”
  Overall, this paper adds very little in terms of new suggestions for moving the field forward beyond what has been covered previously. There are already numerous opinions, reviews, and reports on the application of ML in hydrologic science and modeling, and the paths forward with integration of process-based and ML models. Below is a list of examples of such papers going back several years. None (other than Nearing et al. (2020) of these are cited in this manuscript.
  Painter and Destouni (2026) https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2026WR043509
  
  Yan et al. (2026): https://www.sciencedirect.com/science/article/pii/S1674237026000025
  
  Zhi et al. (2024) https://www.nature.com/articles/s44221-024-00202-z
  
  Varadharajan et al. (2022): https://onlinelibrary.wiley.com/doi/abs/10.1002/hyp.14565
  
  Xu and Liang (2021) https://wires.onlinelibrary.wiley.com/doi/abs/10.1002/wat2.1533
  
  Nearing et al. (2020) https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2020WR028091
  
  Fleming and Gupta (2020) https://physicstoday.aip.org/features/the-physics-of-river-prediction
  
  Sun and Scanlon (2019) https://iopscience.iop.org/article/10.1088/1748-9326/ab1b7d
  Response: Thank you for providing this valuable comment on how our manuscript can be improved and pointing us to relevant references. In response, we have further clarified the novelty and significance of our perspective. Specifically, a paragraph in section 1 expands on the intention and significance of our perspective for the field of hydrology (lines 67-76).
  “The goal of this perspective is to highlight the key challenges that currently hinder the integration of ML within some current operational settings, despite their researched promise of improved predictive accuracy. The focus is not to discuss the adequacy of ML methods relative to conceptual or physics-based models, as such comparisons have been extensively explored in the literature (Painter and Destouni, 2026; Yan et al., 2026; Kok Poh et al., 2022; Xu and Liang, 2021a; Sun and Scanlon, 2019). Rather, we frame this perspective around how organizations that depend on operational hydrologic models should navigate modernization decisions in the era of rapidly advancing ML capabilities, particularly when complete replacement by ML is impractical, undesirable, or infeasible. We specifically highlight the need for organizations to systematically evaluate opportunities for retaining, revamping, reconciling, or replacing existing modelling systems with ML-based approaches. In doing so, we shift the focus from model competition toward defining practical ML pathways for integration within applied model workflows.”
  Additional revisions throughout the manuscript, including those highlighted in our responses above, further reinforce our framing of how organizations that rely on operational hydrologic models can navigate modernization decisions in the era of rapidly advancing ML capabilities.
  Minor Comments
  Table 1 is not useful or comprehensive as a list of process models. There is a huge number of process models used in hydrology (see the extensive list of hydrologic models maintained by the CSDMS https://csdms.colorado.edu/wiki/Hydrological_Models). Alternatively, Figure 1 in Yan et al. (2026) provides a useful overview of the trajectory of hydrologic model development. The main point to make is that there is a huge diversity of process models at varying levels of complexity, and the opportunities for integration with AI can vary a lot depending on the model type. This paper could explore that concept further.
  Response: Thank you for this comment. We intended to have Table 1 provide illustrative examples of how some models are used operationally. We have decided to remove the table within the manuscript as it does not get this point across well. The second part of your comment (addressing diverse model complexity) is a well-made point. We acknowledge that the diversity of model structure, complexity, and use impacts the ultimate integration technique chosen for AI. Within section 2, lines 93-95 highlight how existing models are utilized for diverse decisions:
  “…applied model workflows often represent multi-variable dynamics at varying temporal and spatial time scales which facilitate decision making. For example, groundwater management often requires spatially distributed estimates of hydraulic heads, recharge rates, and groundwater–surface water interactions. Similarly, flood risk assessments depend not only on streamflow predictions, which ML has gotten good at (Song et al., 2025), but also on water depths, velocities, inundation extents, and infrastructure impacts. Many applications further involve evaluating future climate, land-use, infrastructure, or management scenarios for which observational analogues are limited or unavailable. Additional challenges arise from pervasive human modifications to hydrologic systems, including irrigation withdrawals, urban drainage networks, water transfers, and evolving land-use practices.”
  Additionally, we have added a sentence emphasizing how model type and use highly influences the framework within section 4, lines 295-297:
  “It is to be noted that relevance and implementation of individual steps in the roadmap may vary by application, the type of legacy model being used, current limitations in the system, the modification done with ML, and the complexity of the chosen adjustment.”
  The workflow sections are very short, lacking depth, and should be expanded substantially, as the primary contribution of the paper. The integration objectives and criteria for choosing an ML approach are big topics that can be elaborated on further.
  Response: Thank you for this response. We have edited the manuscript to further aid organizations using applied models in implementing ML modules. In addition to the paragraph clarifying ideal ML approach operational and institutional constraints and criteria (see major comment 1), we have added evaluatory questions within section 4, step 1 (lines 272-284), for users to identify an appropriate ML approach.
  “4.1 Step 1: Assessment of organizational objectives and constraints
  
  Firstly, organizations should evaluate organizational objectives, operational requirements, and institutional constraints to guide their selection of ML implementation technique. Key questions to consider include: (i) does the existing model adequately meet operational needs? (ii) Which limitations, when addressed, would improve the model’s utility and/or effectiveness? (iii) Are regulatory approvals, established workflows, or stakeholder expectations tied to the existing modeling framework? (iv) Does prior research demonstrate substantial improvements through ML for some or all of the model’s intended applications? (v) Are sufficient financial, computational, and personnel resources available to support the partial or complete replacement of existing model workflows? Finally, (iv) would the intended implementation technique be sustainable over time? After carefully evaluating these questions, the ideal ML integration technique can be selected, i.e., retention, replacement, reconciliation, or revamping. As an example, an agency may recognize their calibration process limiting SAC-SMA potential. They may find that regulatory guidelines entrench SAC-SMA usage within their organization, despite the potential improvement in model effectiveness with a ML model. They have the desired financial and computational resources and could invest in workflow training over time. They may decide to pursue the revamping approach to meet organizational regulatory requirements while increasing model capabilities.”
  To further define integration objections, we have broadened section 4.2 and provided more examples of possible objectives (lines 285-295).
  “4.2 Step 2: Set ML integration objectives
  
  After identifying the desired ML integration technique, ML integration objectives can be established. If retention is selected, the workflow can be pre-eminently stopped and evaluated again in the future should organizational objectives or limitations change. Otherwise, companies selecting reconciliation, revamping, or replacement should assess and create benchmarks describing the successful implementation of their ML technique. Possible objectives to consider defining include the following: desired accuracy increase or limitation decrease, increased degree of worker comfortability or understanding, decrease in workflow time, and more. Extending the SAC-SMA calibration example, the agency may want to address both calibration accuracy and parameter transferability or focus on only one of these objectives based on operational priorities. For calibration accuracy, they may desire to increase accuracy by 20% while maintaining equivalent levels of operator understanding. For parameter transferability, they may define their objective as a 10% increase in spatial generalization capabilities for nearby catchments.”
  The paper lacks sufficient referencing for an opinion piece. For example, it includes only three references to single-variable machine learning (ML) models, despite the existence of hundreds of publications demonstrating ML model skill across a diverse range of variables. As previously mentioned, numerous review papers that cite hundreds of these studies are available and should be included.
  Response: Thank you for this comment. As suggested, more references have been added, in particular review articles in sections 1 and 2. These include the following and more:
  Yu-fei, Y., Han-xiao, L., Shu, X., Qiong-lin, W., Yu-hui, Y., Qing-qing, C., Chen-yang, W., & Tian-ling, Q. (2026). Advances in coupling machine learning with hydrological simulation: A review. Water Science and Engineering, 19(1), 1-10. https://doi.org/https://doi.org/10.1016/j.wse.2026.01.002
  
  Xu, T., & Liang, F. (2021). Machine learning for hydrologic sciences: An introductory overview. WIREs Water, 8(5), e1533. https://doi.org/https://doi.org/10.1002/wat2.1533
  
  Sun, A. Y., & Scalon, B. R. (2019). How can Big Data and machine learning benefit environment and water management: a survey of methods, applications, and future directions. Environmental Research Letters, 14(7). https://doi.org/10.1088/1748-9326/ab1b7d
  
  Kok Poh, W., Min Yan, C., Chai Hoon, K., Yuk Feng, H., & Woon Chan, C. (2022). Applications of deep learning in water quality management: A state-of-the-art review. Journal of Hydrology, 613, 128332. https://doi.org/https://doi.org/10.1016/j.jhydrol.2022.128332
  
  Additional references regarding single-variable machine learning (ML) models that have now been added include:
  Ghimire, S., Yaseen, Z. M., Farooque, A. A., Deo, R. C., Zhang, J., and Tao, X.: Streamflow prediction using an integrated methodology based on convolutional neural network and long short-term memory networks, Scientific Reports, 11, https://doi.org/10.1038/s41598-021-96751-4, 2021.
  
  Malekzadeh, M., Kardar, S., and Shabanlou, S.: Simulation of groundwater level using MODFLOW, extreme learning machine and Wavelet-Extreme Learning Machine models, Groundwater for Sustainable Development, 9, https://doi.org/10.1016/j.gsd.2019.100279, 2019.
  
  Kratzert, F., Klotz, D., Shalev, G., Klambauer, G., Hochreiter, S., and Nearing, G.: Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets, Hydrol. Earth Syst. Sci., 23, 5089-5110, https://doi.org/10.5194/hess-23-5089-2019, 2019.
  
  Tsai, W.-P., Feng, D., Pan, M., Beck, H., Lawson, K., Yang, Y., Liu, J., and Shen, C.: From calibration to parameter learning: Harnessing the scaling effects of big data in geoscientific modeling, Nature Communications, 12, https://doi.org/10.1038/s41467-021-26107-z, 2021.
  
  Song, Y., Bindas, T., Shen, C., Ji, H., Knoben, W. J. M., Lonzarich, L., Clark, M. P., Liu, J., van Werkhoven, K., Lamont, S., Denno, M., Pan, M., Yang, Y., Rapp, J., Kumar, M., Rahmani, F., Thébault, C., Adkins, R., Halgren, J., Patel, T., Patel, A., Sawadekar, K. A., and Lawson, K.: High-Resolution National-Scale Water Modeling Is Enhanced by Multiscale Differentiable Physics-Informed Machine Learning, Water Resources Research, 61, 10.1029/2024WR038928, 2025.
  
  Fleming, S. W., Garen, D. C., Goodbody, A. G., MyCarthy, C. S., and Landers, L. C.: Assessing the new Natural Resources Conservation Service water supply forecast model for the American West: A challenging test of explainable, automated, ensemble artificial intelligence, Journal of Hydrology, 602, https://doi.org/10.1016/j.jhydrol.2021.126782, 2021.
  
  Nevo, S., Morin, E., Gerzi Rosenthal, A., Metzger, A., Barshai, C., Weitzner, D., Voloshin, D., Kratzert, F., Elidan, G., Dror, G., Begelman, G., Nearing, G., Shalev, G., Noga, H., Shavitt, I., Yuklea, L., Royz, M., Giladi, N., Peled Levi, N., Reich, O., Gilon, O., Maor, R., Timnat, S., Shechter, T., Anisimov, V., Gigi, Y., Levin, Y., Moshe, Z., Ben-Haim, Z., Hassidim, A., and Matias, Y.: Flood forecasting with machine learning models in an operational framework, Hydrol. Earth Syst. Sci., 26, 4013-4032, 10.5194/hess-26-4013-2022, 2022.
  
  We sincerely thank the reviewer again for their time and thoughtful comments. Their feedback certainly helped us strengthen and improve the manuscript.
  
  Citation: https://doi.org/10.5194/egusphere-2026-583-AC2

Delanie Williams, Mukesh Kumar, Katie van Werkhoven, Martyn Clark, Christopher Wilson, and Paul Miller

Viewed

Total article views: 1,899 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
932	894	73	1,899	92	146

HTML: 932
PDF: 894
XML: 73
Total: 1,899
BibTeX: 92
EndNote: 146

Views and downloads (calculated since 19 Feb 2026)

Month	HTML	PDF	XML	Total
Feb 2026	452	323	32	807
Mar 2026	284	302	30	616
Apr 2026	89	90	1	180
May 2026	85	86	6	177
Jun 2026	9	25	2	36
Jul 2026	13	68	2	83

Cumulative views and downloads (calculated since 19 Feb 2026)

Month	HTML	PDF	XML	Total
Feb 2026	452	323	32	807
Mar 2026	284	302	30	616
Apr 2026	89	90	1	180
May 2026	85	86	6	177
Jun 2026	9	25	2	36
Jul 2026	13	68	2	83

Viewed (geographical distribution)

Total article views: 1,827 (including HTML, PDF, and XML) Thereof 1,827 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 16 Jul 2026

Short summary

Machine learning (ML) models have potential within water resource management to be used in place of traditional models. However, there is hesitance for many agencies and institutions to pursue ML models, due to existing familiarity, reliability, interpretability, resources, and workflows. Thus, instead of opting for complete retention of traditional models, or complete replacement by ML models, this perspective argues for the selective integration of ML within existing models.


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