Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper investigates the neuro-fuzzy control for intelligent irrigation based on the Internet of Things (IoT). The authors combine fuzzy control, Artificial Neural Network (ANN) estimation, and stability theory, and conduct short-term validation in an actual orchard, achieving a water-saving effect of 28%. The structure of the paper is generally complete, but the following issues need to be addressed:

1. The introduction occupies a relatively large portion, with some content (such as the literature review) overlapping with subsequent sections. It is recommended to streamline the literature review in Chapter 2.
2. The soil moisture model is overly simplified, neglecting soil heterogeneity, the nonlinearity of root water uptake, and the hysteresis effect of deep percolation. It is advisable to discuss the limitations of the model assumptions.
3. The paper mentions that 17 fuzzy rules are based on "expert knowledge." What is the basis for acquiring this knowledge? It is recommended to supplement or provide an analysis of the rule design. Additionally, explain the difference between Takagi-Sugeno (T-S) fuzzy systems and Fuzzy Logic Systems (FLSs) (as referenced in "Enhanced state-constrained adaptive fuzzy exact tracking control for nonlinear strict-feedback systems").
4. The ANN used for ET estimation is trained with only 54 samples, which is a small sample size with a short time span. Have considerations been given to the potential overfitting phenomenon that may arise?
5. Theorem 1 requires the fuzzy control law to satisfy the sector condition (12). How does this method differ from the Lipschitz continuity condition (as seen in "Robust Locomotion Policy With Adaptive Lipschitz Constraint for Legged Robots")?
6. Comparing only with "fixed-schedule irrigation" fails to fully demonstrate the advantages of the proposed intelligent method. It is recommended to add comparisons with other intelligent methods, such as "Towards Precision Agriculture: IoT-enabled Intelligent Irrigation Systems Using Deep Learning Neural Network."
7. Figure 3 has low resolution, and Figure 6 lacks key coordinate annotations. It is recommended to redraw high-resolution figures and supplement them with elements such as contour lines or dynamic simulation trajectories to enhance readability.

Author Response

Dear Reviewer,

We sincerely thank you for your careful reading of the manuscript and for your constructive comments and suggestions. We believe that addressing these points has significantly improved the quality, clarity, and scientific rigor of the paper. Our detailed responses are provided below.

Comment 1.
The introduction occupies a relatively large portion, with some content overlapping with subsequent sections. It is recommended to streamline the literature review in Chapter 2.

Response:
We thank the reviewer for this observation. The Introduction and Literature Review sections have been carefully revised to reduce overlap. General background material and duplicated references have been removed from the Introduction and consolidated into Section 2. As a result, the Introduction is now more focused on motivation, problem formulation, and contributions, while the Literature Review provides a clearer and more concise synthesis of related work.

Comment 2.
The soil moisture model is overly simplified, neglecting soil heterogeneity, nonlinear root water uptake, and hysteresis effects.

Response:
We agree with the reviewer that real soil–plant systems exhibit significant heterogeneity and nonlinearities. The proposed model is intentionally simplified to ensure real-time implementability on low-cost IoT hardware. To address this concern, we have added a dedicated discussion of model limitations in Section 5 (Limitations and Discussion), explicitly acknowledging the neglected effects and clarifying that they are partially compensated by (i) online RLS parameter adaptation and (ii) the ANN-based evapotranspiration estimator. This clarification strengthens the transparency and applicability of the modeling assumptions.

Comment 3.
The paper mentions that 17 fuzzy rules are based on “expert knowledge.” Please clarify how this knowledge was acquired and explain the difference between Takagi–Sugeno (T–S) fuzzy systems and Fuzzy Logic Systems (FLSs).

Response:
We appreciate this important request for clarification. The manuscript has been revised to explicitly state that the expert knowledge originates from standard irrigation heuristics used by agronomists (soil moisture thresholds, temperature-driven ET behavior, and radiation influence), combined with empirical tuning during preliminary field trials. Furthermore, a clear explanation distinguishing Mamdani-type Fuzzy Logic Systems (FLSs) from Takagi–Sugeno (T–S) fuzzy systems has been added. We also clarify that the practical controller is Mamdani-type, while its theoretical analysis relies on an equivalent zero-order T–S representation under centroid defuzzification, which enables stability analysis without altering the implemented logic.

Comment 4.
The ANN used for ET estimation is trained with only 54 samples, which raises concerns about overfitting.

Response:
We thank the reviewer for highlighting this point. The limited dataset reflects the constraints of real field experiments rather than an attempt to build a large-scale predictive ANN. To mitigate overfitting, the ANN architecture is deliberately lightweight (single hidden layer, small neuron count), and it is used strictly as an auxiliary estimator rather than as the main decision-making mechanism. This role clarification and a discussion of overfitting risk and mitigation strategies have been added to Section 4.4 and the Discussion section.

Comment 5.
Theorem 1 requires a sector condition. How does this differ from Lipschitz continuity conditions used in other works?

Response:
We thank the reviewer for this technically insightful question. The manuscript now clarifies that the sector condition is a structural property of the fuzzy control law that guarantees bounded monotonic behavior, whereas Lipschitz continuity is a regularity condition on system nonlinearities. The sector condition is weaker and more suitable for fuzzy controllers, while Lipschitz constraints are often imposed on learning-based policies. This distinction and its relevance to the stability proof have been explicitly explained in Section 5.

Comment 6.
Comparison only with fixed-schedule irrigation is insufficient. Comparisons with other intelligent methods are recommended.

Response:
We agree that broader comparisons are valuable. Due to hardware, deployment, and data constraints, real-time field comparison was limited to fixed-schedule irrigation. However, we have expanded the Discussion section to include a qualitative and quantitative comparison with representative intelligent irrigation approaches from the literature, including deep learning–based and IoT-enabled methods. This contextualizes the proposed approach within the current state of the art while maintaining experimental integrity.

Comment 7.
Figure 3 has low resolution, and Figure 6 lacks key coordinate annotations.

Response:
We appreciate this practical suggestion. All figures have been revised to improve resolution and readability. Figure 3 has been redrawn at higher resolution, and Figures 6–8 now include labeled axes, units, and clearer surface annotations to better convey the controller behavior.

 

Once again, we thank the reviewer for the insightful and constructive feedback, which has helped us substantially improve the manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript provides some heuristic approach for IoT based ANFIS smart irrigation system. The results presented are not compared with any alternate existing methodology. 

All 42 references are very briefly described in just section 1 and 2. Which of these references resulted in the research gap and which of these references are used to develop the materials and methodology presented in the manuscript are not cited. Entire mathematics presented in section 4 and the theorems and their proof presented in section 5 is not supported by citation of any references. The assumptions made in section 4 and section 5 are not justified. The theorems presented in section 5 whether are taken from any literature or developed by authors is not made clear. Proofs of all Lemma's and theorems in Section 5 are very qualitative and heuristic based which is not validated by any rigorous mathematical background. 

The results provided are not supported by any experimental data and given superficially as 0.03 and 28%, which is not be validated satisfactorily.The methodology adapted is not validated by any state of the art alternate methodologies, implementation of the methodology on real time system is not described satisfactorily and how weights of ANN, number of hidden nodes arrived at etc are not presented comprehensively.

Overall the manuscript in present form is presenting the idea superficially without necessary technical/mathematical validation. 

Author Response

Dear Reviewer,

We thank you for your thorough and critical evaluation of the manuscript. We have carefully revised the paper to address all concerns raised. Below, we provide a detailed, point-by-point response outlining the specific changes made to improve the technical rigor, clarity, and validation of the proposed methodology.

Comment 1

The manuscript provides some heuristic approach for IoT based ANFIS smart irrigation system. The results presented are not compared with any alternate existing methodology.

Response:
We respectfully note that the proposed framework is not a purely heuristic ANFIS-based approach. The manuscript has been revised to clearly emphasize that the irrigation controller is a Mamdani-type fuzzy system with an equivalent Takagi–Sugeno (TS) representation used exclusively for theoretical analysis. To address the concern regarding comparison with alternative methodologies, a new subsection “Comparison with Other Intelligent Irrigation Methods” has been added to the Discussion section. This subsection positions the proposed method against deep learning–based controllers, ANFIS/metaheuristic approaches, and model predictive control (MPC), highlighting differences in interpretability, computational requirements, and availability of formal stability guarantees.

Comment 2

All 42 references are very briefly described in just section 1 and 2. The research gap and methodological foundations are not clearly linked to the cited literature.

Response:
This concern has been addressed by strengthening the traceability between the reviewed literature and the proposed methodology. A new summary and research gap paragraph has been added at the end of the Literature Review, explicitly distinguishing between heuristic fuzzy controllers, data-driven ANN approaches, and control-theoretic TS/LMI-based methods. Furthermore, the beginning of the Related Work section now clearly positions the present work relative to representative references, explicitly stating how the proposed framework builds upon and extends existing studies.

Comment 3

The mathematics in Sections 4 and 5 is not supported by references; assumptions are not justified; it is unclear whether the theorems are taken from literature or developed by the authors.

Response:
We have substantially revised Sections 4 and 5 to explicitly ground the mathematical development in established control theory. Canonical references to nonlinear systems and Input-to-State Stability (ISS) theory, Takagi–Sugeno fuzzy modeling, and LMI-based stability analysis have been added at appropriate locations (e.g., Khalil’s Nonlinear Systems, TS fuzzy stability literature). A clarifying statement has been included at the beginning of Section 5, explicitly stating that the theorems are developed by the authors by specializing standard ISS and TS–LMI tools to the proposed soil-moisture and neuro-fuzzy irrigation framework. This makes the origin and scope of the theoretical results unambiguous.

Comment 4

Proofs of lemmas and theorems are qualitative and heuristic, lacking rigorous mathematical background.

Response:
The proofs have been carefully reviewed and clarified in relation to standard nonlinear stability arguments. Additional remarks and citations have been added to show that the Lipschitz assumptions, ISS reasoning, and TS–LMI quadratic stability conditions follow well-established methodologies in the control literature. The proofs are intentionally presented in a concise and self-contained form, consistent with common practice in applied control and cyber-physical systems research, while being firmly anchored in rigorous theoretical foundations.

Comment 5

The results are not supported by experimental data and are reported superficially as 0.03 and 28%.

Response:
We clarify that the reported values correspond to (i) the root-mean-square error (RMSE) obtained from real IoT field data collected over a nine-day experimental period and (ii) measured water savings relative to a fixed-schedule baseline in the same orchard. The experimental setup, IoT architecture, and validation procedure are described in detail in Section 4, and the model-validation results are illustrated in Figure 9. The Discussion section has been revised to explicitly link these quantitative results to the underlying data and experimental protocol.

Comment 6

The methodology is not validated against state-of-the-art methods, and ANN design choices are not explained.

Response:
The manuscript now includes an explicit qualitative comparison with state-of-the-art intelligent irrigation approaches (deep learning, ANFIS, MPC), explaining the trade-offs between data requirements, interpretability, computational complexity, and availability of formal stability guarantees. Regarding the ANN design, Section 4.4 has been expanded to explain the rationale behind the lightweight architecture (single hidden layer, limited number of neurons), emphasizing that the ANN is used strictly as an auxiliary estimator rather than as a primary control policy. This design choice mitigates overfitting risks given the limited dataset and preserves closed-loop stability.

Comment 7

Overall, the manuscript presents the idea superficially without sufficient technical/mathematical validation.

Response:
We respectfully disagree with this assessment after revision. The manuscript now clearly integrates physically motivated modeling, interpretable fuzzy control, ANN-based estimation, and mathematically certified stability analysis within a single IoT-deployable framework. The added literature traceability, explicit theoretical grounding, clarified assumptions, and strengthened experimental discussion collectively ensure that the proposed approach is technically sound, mathematically justified, and practically validated.

 

We sincerely thank the reviewer for the constructive criticism, which has significantly improved the clarity, rigor, and presentation of the manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper proposes an IoT-enabled irrigation control framework that couples a physically motivated discrete-time soil-moisture model identified online via RLS with an interpretable Mamdani fuzzy controller, then derives an equivalent TS form to enable ISS/LMI-based stability certification. It further augments the controller with lightweight ANN modules for evapotranspiration estimation and slow membership-function adaptation, and reports field deployment results showing <2% tracking error and about 28% water savings versus fixed scheduling.
Major revision is recommended. 
1. Although the literature review has been separated into its own section, the reviewer feels it is not sufficiently aligned to clearly highlight the paper’s contributions.
2. The literature review related to fuzzy control, Takagi–Sugeno (T-S) models, and LMI is still not solid or accurate.
3. The fuzzy rule construction in Section 4.3 is overly simplistic; it lacks a stability proof and does not include any LMI-based analysis.
4. Similarly, the neural network part is also very basic and lacks depth.
5. The conversion from the T-S representation to the Mamdani fuzzy system is not sufficiently justified to guarantee stability.
6. The paper does not provide standard LMI conditions, and the sub-block matrix components are not clearly defined based on the T-S model.
7. The results section includes no figures or visual outputs, so the research outcomes are unclear; the results need to be presented more concretely and explicitly.

Author Response

Dear Reviewer,

We sincerely thank you for the detailed and technically informed review of our manuscript and for recognizing the overall structure and objectives of the proposed framework. In response to your comments, we have substantially revised the manuscript to improve the alignment of the literature review, strengthen the theoretical foundations, clarify the role of the fuzzy and neural components, and present the experimental results more explicitly. Our detailed responses are provided below.

Comment 1

Although the literature review has been separated into its own section, it is not sufficiently aligned to clearly highlight the paper’s contributions.

Response:
We agree with this observation and have revised the literature review to improve its alignment with the paper’s contributions. A dedicated summary and research gap paragraph has been added at the end of Section 2, explicitly distinguishing between heuristic fuzzy irrigation controllers, data-driven ANN approaches, and control-theoretic TS/LMI-based methods. Furthermore, the beginning of Section 3 now clearly positions the present work relative to representative studies, highlighting how it integrates physically grounded modeling, interpretable Mamdani fuzzy control, ANN-based estimation, and mathematically certified stability analysis within a single framework.

Comment 2

The literature review related to fuzzy control, Takagi–Sugeno models, and LMI is still not solid or accurate.

Response:
The manuscript has been strengthened by explicitly anchoring the discussion of Takagi–Sugeno fuzzy models and LMI-based stability tools to canonical references in fuzzy control and nonlinear systems theory. Section 2.1 and the opening of Section 5 now explicitly cite established TS–LMI stability results and clarify their relevance to the present irrigation-control problem. This ensures that the use of TS modeling and LMI-based quadratic stability certification is grounded in well-established theory rather than presented as an ad hoc construction.

Comment 3

The fuzzy rule construction in Section 4.3 is overly simplistic and lacks stability proof and LMI-based analysis.

Response:
We clarify that the Mamdani fuzzy rule base in Section 4.3 is intentionally designed to be simple and interpretable, reflecting agronomic heuristics and practical deployment constraints. Stability is not established directly at the level of individual Mamdani rules. Instead, stability guarantees are derived by exploiting the equivalent zero-order Takagi–Sugeno representation of the Mamdani controller under centroid defuzzification. This equivalence is explicitly described in Section 5.1, and the resulting TS model is analyzed using LMI-based quadratic stability conditions in Section 5. The stability proof therefore applies to the implemented controller through this mathematically established equivalence.

Comment 4

The neural network part is very basic and lacks depth.

Response:
The ANN component is intentionally lightweight by design. Its role is limited to evapotranspiration estimation and slow adaptation of membership-function parameters, and it is not used as a primary control policy. This design choice reflects the constraints of real-time execution on resource-limited IoT hardware and the need to preserve interpretability and stability. Section 4.4 has been expanded to explain the rationale behind the ANN architecture (single hidden layer, limited number of neurons, ReLU activation), and to clarify how this design mitigates overfitting risks while supporting the stability analysis presented in Section 5.

Comment 5

The conversion from the T–S representation to the Mamdani fuzzy system is not sufficiently justified to guarantee stability.

Response:
We have clarified that the practical controller is Mamdani-type, while the Takagi–Sugeno representation is used exclusively for theoretical analysis. Section 5.1 now explicitly explains the equivalence between a Mamdani controller with centroid defuzzification and a zero-order TS fuzzy model. Since the TS representation yields a weighted-average form with constant consequents, the ISS and LMI-based stability results derived for the TS model apply directly to the implemented Mamdani controller. This justification is now stated explicitly to avoid ambiguity.

Comment 6

The paper does not provide standard LMI conditions, and the sub-block matrix components are not clearly defined based on the T–S model.

Response:
The LMI conditions presented in Theorem 2 follow standard TS fuzzy quadratic stability formulations. To improve clarity, we have added explicit references to established TS–LMI methodologies and clarified the structure of the local closed-loop subsystems and the role of the weighting functions. The LMI formulation is presented in a compact form consistent with applied control literature, while being explicitly grounded in standard TS fuzzy stability results.

Comment 7

The results section includes no figures or visual outputs, so the research outcomes are unclear.

Response:
We respectfully note that the revised manuscript includes multiple figures illustrating the system architecture, membership functions, fuzzy control surfaces, and experimental validation results. In particular, Figure 9 explicitly compares measured and simulated soil-moisture trajectories, supporting the reported RMSE value, and Figures 6–8 visualize the fuzzy control behavior. The Results and Discussion sections have been revised to explicitly link these figures to the reported performance metrics (tracking error below 2% and approximately 28% water savings), ensuring that the experimental outcomes are presented clearly and concretely.

 

We thank the reviewer again for the constructive feedback, which has significantly improved the rigor, clarity, and presentation of the manuscript.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The revised manuscript contains subjective/qualitative clarifications for many queries raised by the reviewer. These clarifications improved theoretical presentation in the manuscript. However, some of following concerns still are unanswered justifiably/lacks clarity and some comments to improve the quality of manuscript are presented below.

Statement given in line 389-392 how interpreted from Fig 6-8 is unclear.

section 4.3 and 4.4 no supporting literature cited. Whether the content presented is original work done by authors or referred from existing literature is unclear.

Section 5.1 content presented in line 593, 602-603 not given any reference or citation.

Statement given in line 643-644 regarding 28% reduction. It is unclear how the number arrived at using Fig 6-8. Necessary experimental data be provided or methodology of arriving at this number be presented.

Section 6.2 line 637-638 It is unclear how the results arrived at as no supporting experimental results presented in the manuscript.

section 7.3 line 659-660 Presented results not supported by experimental data.

Section 8.3 presentation is very superficial as no supporting references cited.

Following are a few corrections recommended.

parameter D(x) and d(x) used interchangeably at many places. Eq (1), (6), (9),  lines 312, 472, 474, 480 etc. Suggested to be consistent. 

Fig 6-8, Significance of different colors in 3 D plot and how they are interpreted is not explained. 

Fig 3,4, 6,7,8 axis labels are not readable. Suggested to increase font size.

line 404 ISS stability word stability could be deleted.

line 681 check the spelling of vine of wine

 

Authors are suggested to incorporate above listed comments to make otherwise theoretically strong manuscript to be supported with correspondingly strong experimental validation.

Author Response

Thank you very much for your thorough review and constructive remarks. We truly appreciate the effort invested in evaluating our manuscript and we have carefully revised the paper accordingly. Below we provide detailed point-by-point answers.

 

1. Statement in lines 389–392 is unclear from Figs. 6–8

 

Thank you for the comment. We have clarified the interpretation in Section 4.3 by explicitly explaining how the graphical surfaces correspond to the linguistic control rules and the nonlinear controller behavior.

 

2. Sections 4.3 and 4.4 contain no supporting literature

 

Thank you very much for your remark. Additional references have now been included in Sections 4.3 and 4.4 to clearly position our contribution relative to existing fuzzy and TS-based irrigation control approaches. We have also explicitly stated which parts are original contributions of the authors.

 

3. Section 5.1 (lines 593, 602–603) lacks references

 

Thank you for the observation. The text in Section 5.1 has been revised and appropriate references have now been added to support the presented methodology.

 

4. The 28% reduction statement is unclear

 

Thank you for this important remark.

Section 6.2 has been substantially revised. We now: explicitly present the water consumption computation methodology, provide clear comparison between the proposed controller and fixed-schedule irrigation, include supporting numerical values and explanation of how the 28% saving was derived from the experimental results.

This clarification ensures transparency and reproducibility of the reported savings.

 

5. Section 6.2 (lines 637–638) lacks supporting experimental results

 

Thank you for the comment. As mentioned above, Section 6.2 has been significantly updated. We now provide explicit experimental validation details and data-based justification of the obtained results.

 

6. Section 7.3 (lines 659–660) results not supported by data

 

Thank you very much for your remark. Section 7.3 has been revised to better connect the discussed results with the experimental findings and to avoid any unsupported claims.

 

7. Section 8.3 is superficial and lacks references

 

Thank you for pointing this out. Section 8.3 has been rewritten to provide deeper discussion and is now supported with relevant citations, placing our work in the broader context of intelligent irrigation and control theory.

 

Corrections and Technical Improvements

 

D(x) vs d(x) notation inconsistency

Thank you for the recommendation. All occurrences have been checked and corrected to ensure consistent notation throughout the manuscript.

 

Figures 6–8: color meaning not explained

Thank you for the remark. The colors are predefined in the used software product (MATLAB, Fuzzy logic library).

 

Figures 3, 4, 6, 7, 8 axis readability

We acknowledge this important suggestion. The figures have been regenerated with increased resolution and enlarged axis labels to improve readability.

 

Line 404 – “ISS stability” wording

Corrected.

 

Line 681 spelling (“vine/wine”)

Corrected.

English Quality

Thank you for your positive evaluation of the English quality.

 

Thank you once again for your valuable comments. They helped us significantly improve the clarity, rigor, and presentation of our manuscript, strengthen the theoretical contribution, and provide clearer experimental justification.

Reviewer 3 Report

Comments and Suggestions for Authors

The author has significantly improved the comments. 

Author Response

We sincerely thank the reviewer for the positive evaluation and appreciation of the improvements made. The provided comments were extremely valuable for strengthening the clarity, rigor, and overall quality of the manuscript.

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Reviewer appreciate the efforts by authors to address many of the queries satisfactorily. However, reviewer could find some discrepancy in the revised manuscript and suggest to correct the same.

Fig 6 and Fig 8 are duplicated, with different caption. Suggested to  correct appropriately. Also, the axis marking in Fig 6-7 are not consistent. Humidity is marked at front in fig 7 and left on Fig 6. Fig 7 could be convincingly comprehended, with regards to explanation in line 409-415 and 702-703 but not so from Fig 6. In fig 6 it seems high pump output with low temperature, if the reviewer interpreted correctly. Hence to avoid confusion, it is suggested to add a table with quantitative values of pump output as a function of humidity, radiation, and temperature to support the 3 D plot. More over significance of different coloring in 3D plot be mentioned for more clarity.

Further is it not the performance with regard to radiation and temperature be identical. Justify why these two variables are explicitly taken separately in the analysis.

Further section 7.3, line 716-720, 724-725 claim made is not supported by evidence in the manuscript. Either add supporting experimental/simulation results to validate the claim or avoid claim without supporting simulation/experimental result.

Authors are suggested to look into above queries and provide explanation and make necessary corrections.

Author Response

Comments to the Authors

Reviewer appreciate the efforts by authors to address many of the queries satisfactorily. However, reviewer could find some discrepancy in the revised manuscript and suggest to correct the same.

• Fig 6 and Fig 8 are duplicated, with different caption. Suggested to correct appropriately.
• Also, the axis marking in Fig 6-7 are not consistent. Humidity is marked at front in fig 7 and left on Fig 6.
• Fig 7 could be convincingly comprehended, with regards to explanation in line 409-415 and 702-703 but not so from Fig 6. In fig 6 it seems high pump output with low temperature, if the reviewer interpreted correctly. Hence to avoid confusion, it is suggested to add a table with quantitative values of pump output as a function of humidity, radiation, and temperature to support the 3 D plot. More over significance of different coloring in 3D plot be mentioned for more clarity.
• Further is it not the performance with regard to radiation and temperature be identical. Justify why these two variables are explicitly taken separately in the analysis.
• Further section 7.3, line 716-720, 724-725 claim made is not supported by evidence in the manuscript. Either add supporting experimental/simulation results to validate the claim or avoid claim without supporting simulation/experimental result.

Authors are suggested to look into above queries and provide explanation and make necessary corrections.

Comments on the Quality of English Language

The English is fine and does not require any improvement.

 

To Reviewer 2:

            Thank you very much for your review and valuable remarks.

- Fig 6 and Fig 8 are duplicated, with different caption. Suggested to  correct appropriately.

- Thank you very much for the remark. Fig 8 has been replaced accordingly.

- Also, the axis marking in Fig 6-7 are not consistent. Humidity is marked at front in fig 7 and left on Fig 6.

- Thank you very much for your comment. The axis labeling and positioning is predefined in the used software product (MATLAB, Fuzzy logic library).

- Fig 7 could be convincingly comprehended, with regards to explanation in line 409-415 and 702-703 but not so from Fig 6. In fig 6 it seems high pump output with low temperature, if the reviewer interpreted correctly. Hence to avoid confusion, it is suggested to add a table with quantitative values of pump output as a function of humidity, radiation, and temperature to support the 3 D plot. More over significance of different coloring in 3D plot be mentioned for more clarity.

- Thank you very much for your comment. Additional text was included to explain the output results (line 397-408). The graphics are being generated by MATLAB based on the linguistic rules set up. The colors, axis labeling and values are predefined in the product itself.

- Further is it not the performance with regard to radiation and temperature be identical. Justify why these two variables are explicitly taken separately in the analysis.

- Thank you very much for the remark. Additional text has been included to explain the reason behind selecting the three input parameters (line 340-345).

- Further section 7.3, line 716-720, 724-725 claim made is not supported by evidence in the manuscript. Either add supporting experimental/simulation results to validate the claim or avoid claim without supporting simulation/experimental result.

- We thank the reviewer for the careful reading of the manuscript and for this important remark. We agree that the claims originally stated in Section 7.3 were not sufficiently supported by explicit experimental or simulation evidence presented in the manuscript. In the revised version, we have addressed this issue by clarifying the scope and nature of the reported neuro-fuzzy improvements.

Specifically, we now explicitly state that the reported performance gains of the neuro-fuzzy extension are based on controlled simulation studies using the identified soil-moisture model, rather than on additional field experiments. The claims have been reformulated to be indicative and qualitative, and any statements implying full experimental validation have been removed. Furthermore, we clearly indicate that extended multi-season experimental validation of the neuro-fuzzy controller is part of future work.

These revisions ensure that all claims in Section 7.3 are fully consistent with the evidence provided in the manuscript and avoid unsupported quantitative assertions. We thank the reviewer for helping us improve the clarity and rigor of the presentation.

 

Comments on the Quality of English Language

The English is fine and does not require any improvement.

- Thank you very much.

 

Thank you very much for your remarks and comments. They were very useful for me to emphasize the main tasks and contributions of the manuscript, and also to focus the attention of the readers on the new and unique elements.