Research on Coupling Knowledge Embedding and Data-Driven Deep Learning Models for Runoff Prediction
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper proposes a runoff prediction model that couples knowledge embedding with data-driven approaches, aiming to establish a dual-driven model. This coupled model breaks the "black box" nature of neural networks, effectively addressing the poor prediction performance of data-driven models at runoff extremes, ensuring that runoff predictions conform to the probability density function of runoff, thereby improving model accuracy. This work is crucial for agricultural decision-making, water resource management, and disaster assessment. I recommend minor revision for this manuscript.
Please see my detailed comments below:
Abstract:
Line 13: please clarify which Data-driven model used for the first occurrence. Be specific in abstract section
Line 22-23: how closely the coupling model aligns with the probability distribution. Please be specific, straightforward, and quantitative here. No vague words/sentences in abstract.
Introduction: introduction section is well-written. No revision needs to be there.
Line 111: Please change title ‘research background' to’ research methodology’
Line 346: How did you initialize the parameters (positions, velocities) of the IPSO algorithm? Please clarify
Line 353: how did you determine the hyperparameters of the IPSO-TCN model.
Line 354: which python module or framework you used for optimizing the particles?
Figure 8 does not clearly demonstrate how domain knowledge be embedded into the loss function to optimize and adjust the model topology or structures. Please re-design it in more clear way, Or maybe add some descriptions to the diagram workflow.
Section 4.4.1 should be moved to the methodology section.
In discussion section, can you discuss the benefits/limitations of your research, compared with others?
Figure 16 is better to be placed in results section.
Please highlight your work’s contribution and significance in the conclusion section.
Comments on the Quality of English Language
The English language is understandable and acceptable for publication.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript titled Research on Coupling Knowledge Embedding and Data Driven Deep Learning Models for Runoff Prediction can be accepted after moderate (Major) revision.
Abstract
In my opinion, this part could be rewritten in one sentence. The authors can better explain why they analyzed this topic. The authors need to better explain why they used these model. If there any other or similar model?
Key words
I think the authors can add one word that explains the research in this study in more detail.
Due to the large number of scientific terms, I recommend that the authors add a section labeled "Abbreviations".
Lines between 29 and 32 can the authors better describe worldwide methods used in analyzes of runoff or water balance? For example, the methods as Lvovich is very good for the fine analyses.
The authors stated that used SWAT model is there any other similar model can be used within this research.
Line 63, how the neural network method can be use in analyses of runoff? Explain better.
I think that in this part of the manuscript the authors must divide methods on numerical, GIS, Remote Sensing and qualitative.
Because the term runoff is strictly connected with the drainage basins and rivers properties and based on all labeled before I highly recommend to the authors to read and cite two valuable references (research).
The recommended references are
- Valjarević, A. GIS-Based Methods for Identifying River Networks Types and Changing River Basins. Water Resour Manage (2024). https://doi.org/10.1007/s11269-024-03916-7.
- Tsegaw, A. T., Skaugen, T., Alfredsen, K., & Muthanna, T. M. (2020). A dynamic river network method for the prediction of floods using a parsimonious rainfall-runoff model. Hydrology Research, 51(2), 146-168. https://doi.org/10.2166/nh.2019.003.
Research background
Fig. 1 Very well presented map
In this part of the manuscript I recommend to the authors to add more about geographical and hydrological features. Especially it is important to explain better river network properties in this plateau.
Also did the meteorological data download from the official and calibrated meteorological stations.
Materials and Methods
Can the authors better explain runoff index?
Standardized Precipitation Index (SPI) the authors try to explain, can the authors better describe how this index varied from region to region.
Eq 1, is this integral has a shape factor or not? Explain better.
Is this Figure 2 (flowchart) present real function of this work and researched area.
Transfer entropy theory
Is there any limit within this theory, explain better?
The entropy of system is always try to be ΔH>0. Can the authors explain how the runoff is connected with the entropy of drainage network system?
Results
How the authors estimated (calculated) drought frequency?
Can the authors better explain two meteorological cycles (30 years, 30 years).
The all maps are overall well drawn.
The section Discussion must be extended and better discussed.
Conclusion
In this section, the authors need to find answers to the following questions?
Why is this research important?
What are the main findings?
The paper has the scientific potential to be published after moderate (major) revision
I recommend a Major revision
Good luck to the authors
Reviewer#2
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsDear Authors,
This manuscript presents a novel approach to runoff prediction in the Yellow River watershed by coupling knowledge embedding with data-driven methods. The study identifies precipitation, sunshine duration, and relative humidity as key factors influencing runoff occurrence. Overall, the manuscript is well-written and suitable for publication. However, the following suggestions may further strengthen the work:
Enhanced Introduction: While the importance of accurate runoff prediction is acknowledged, the introduction could benefit from a stronger connection between natural and/or anthropogenic impacts on the Yellow River's hydrology. This would provide a more comprehensive context for the study's significance.
Study area: Providing detailed information on the watershed's geography, climate, hydrology, and the specific characteristics of the nine hydrological stations selected would enhance understanding of the study's scope and the representativeness of the data.
Discussion section: A more thorough discussion is needed that integrates the study's findings with existing literature. Comparing your results with those of other studies conducted on the Yellow River watershed or similar regions would strengthen the manuscript's contribution to the field.
Conclusion: The manuscript with recommendations for future research directions based on the study's findings would provide valuable insights for further investigation in this study area.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript entitled Research on Coupling Knowledge Embedding and Data-Driven Deep Learning Models for Runoff Prediction, in my opinion, acceptable as presented.
The authors have responded very seriously to all my comments and corrected all typographical and scientific errors.
The authors have now produced much better and more appropriate maps
I therefore recommend acceptance.
Sincerely,
Reviewer#1
Comments on the Quality of English Language
The manuscript entitled Research on Coupling Knowledge Embedding and Data-Driven Deep Learning Models for Runoff Prediction, in my opinion, acceptable as presented.
The authors have responded very seriously to all my comments and corrected all typographical and scientific errors.
The authors have now produced much better and more appropriate maps
I therefore recommend acceptance.
Sincerely,
Reviewer#1
