Enhancing Temperature Data Quality for Agricultural Decision-Making with Emphasis to Evapotranspiration Calculation: A Robust Framework Integrating Dynamic Time Warping, Fuzzy Logic, and Machine Learning
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors obtained a new framework by combining traditional quality control methods with dynamic time warping, fuzzy logic and machine learning, which somewhat improved the reliability of temperature data collected by agrometeorological networks. However, there are still some problems in the article that need to be improved.
- The abstract is too short and the coherence of the last few lines is not strong.
- The introduction should have expressed other people's relevant research to compare with the author's own research, so as to reflect the superiority of the author's own research and make the article more logical.
- The article has not been scrutinized, for example, the same sentence appears twice in line 165, there is an extra decimal point after 8 in line 579, and there are errors in the serial numbers of some of the headings in the article.
- In the 4. Results section of the article, the author presents a complete table of the results shown by the traditional quality control methods, and only a few sentences on the results of his improved DTW-Fuzzy spatial test.
- Table 8 shows a negative improvement for DQI in general, indicating that further improvements are needed.
Moderate editing of English language required.
Author Response
Comments 1: The abstract is too short and the coherence of the last few lines is not strong.
Response 1: Thank you for your valuable feedback. In response to your comment that “the abstract is too short and the coherence of the last few lines is not strong,” we have revised the abstract to provide a more comprehensive summary of the study’s objectives, methods, and contributions
--------------
Comments 2: The introduction should have expressed other people's relevant research to compare with the author's own research, so as to reflect the superiority of the author's own research and make the article more logical.
Response 2: Thak you. for your comment. We have revised Introduction and also recent refereces was added to support this work.
[4] P. Cerlini, L. Silvestri, and M. Saraceni, “Quality control and gap-filling methods applied to hourly temperature observations over Central Italy,” Meteorol. Appl., vol. 27, 2020.
[5] M. Boujoudar et al., “Comparing machine learning algorithms for imputation of missing time series in meteorological data,” Neural Comput. Appl., Dec. 2024, doi: 10.1007/s00521-024-10601-8.
--------------
Comments 3:The article has not been scrutinized, for example, the same sentence appears twice in line 165, there is an extra decimal point after 8 in line 579, and there are errors in the serial numbers of some of the headings in the article.
Respose 3:
Line 165 double sentese "In temperature data, these anomalies occur..." was removed
Line 579 decimal number "8.664" was corrected to "8664"
Also lines 108, 214, 268, 477, 784, 785 references at formation (like [1],[2]) was made as [1], [2]
Equaltion numbering was corrected.
I am so sorry but i still can locate the problem you refer in the serial numbers of some of the headings in the article.
--------------
Comments 4:In the 4. Results section of the article, the author presents a complete table of the results shown by the traditional quality control methods, and only a few sentences on the results of his improved DTW-Fuzzy spatial test.
Response 4:
We thank the reviewer for this valuable comment. In the revised version of the manuscript, we have expanded the Results section to better present the performance of the DTW-Fuzzy spatial test.
Specifically:
-
A new subsection 4.3 "Results of the Proposed DTW-Fuzzy Spatial Test" was added.
-
A dedicated Table 9 was included, summarizing the number and percentage of anomalies detected by the DTW-Fuzzy test for each station.
-
A new Figure X was added, illustrating an example where the DTW-Fuzzy method successfully detected a spatial anomaly not identified by traditional spatial consistency checks.
-
Additional discussion was provided, highlighting the advantages of the DTW-Fuzzy approach in terms of resilience to missing data, seasonal adaptation, and reduced false positives.
--------------
Comments 5:Table 8 shows a negative improvement for DQI in general, indicating that further improvements are needed.
Response 5: Thanks for your comment. The negative DQI is due the ML reconstruction on high data quality measurements. A paragraph to explain technically the issue was added.
"For stations with initially high data quality, such as Ag and Kam, a slight decrease in the overall Data Quality Index (DQI) was observed following machine learning (ML) reconstruction. This effect is explained by the fact that ML models, despite their high predictive accuracy (RMSE 0.40°C–0.66°C), inevitably introduce small deviations relative to the original measurements. When applied to already high-quality data, these minor inaccuracies slightly reduce the accuracy component of the DQI. Given that the DQI formula equally weights completeness and accuracy, this marginal loss offsets the completeness gain. This observation suggests that for datasets with near-perfect initial conditions, a more targeted reconstruction strategy may be preferable, but this will be a future work."
Reviewer 2 Report
Comments and Suggestions for AuthorsThe author proposes a data quality assessment framework for hourly temperature data from a network of six agricultural meteorological stations in the Epirus region of Greece (2015-2023), significantly enhancing the reliability of the dataset and providing strong support for more effective decision-making. However, the paper still has the following shortcomings:
- In Section 1, the existing data quality evaluation methods should be supplemented, including their advantages and disadvantages, and clearly indicate the issues present in current methods. It would help the reader better understand how the proposed data quality assessment framework effectively addresses these issues, particularly in terms of improvements in data completeness, accuracy, and spatial consistency, thereby highlighting the framework’s innovation and practical application value.
- In Section 3.1.2, values that fail the step test are flagged and converted to NaN, and the NaN-marked data are reconstructed using the method described in Section 3.3. However, how is the reliability of the reconstructed data assessed?
- In Section 3.1.4, the author defines Thmin and Thmax as the average of the minimum and maximum temperatures from all meteorological stations as the threshold for the entire network. This does not fully account for the local climate characteristics and geographical differences of each station. The climate variations between different stations may cause these average values to be unsuitable for all stations, especially under extreme weather or specific climatic conditions.
- There is confusion in the chapter numbering, and the author needs to renumber the sections.
- There are duplicate equation numbers, such as two different equations both labeled as (1).
Author Response
Thank you for your valuable feedback.
Comments 1:In Section 1, the existing data quality evaluation methods should be supplemented, including their advantages and disadvantages, and clearly indicate the issues present in current methods. It would help the reader better understand how the proposed data quality assessment framework effectively addresses these issues, particularly in terms of improvements in data completeness, accuracy, and spatial consistency, thereby highlighting the framework’s innovation and practical application value.
Response 1: A new paragraph added (lines 48 -60)
Despite the utility of traditional quality control methods like gross error limits, step tests, and persistence checks, these techniques exhibit several limitations. They primarily focus on identifying obvious anomalies without adequately capturing contextual or spatial inconsistencies, especially when errors appear reasonable in isolation. Furthermore, they often lack mechanisms to deal with extended periods of missing data, introducing gaps that can compromise model reliability. Spatial consistency tests, while promising, are sensitive to missing values and seasonal variations, reducing their effectiveness in automated pipelines. These challenges underscore the need for a more robust framework that not only enhances data completeness and accuracy but also ensures spatial coherence across networks. In this work, by integrating Dynamic Time Warping (DTW), Fuzzy Logic, and machine learning-based reconstruction, we address these limitations, offering an innovative and practical solution for the comprehensive assessment and improvement of temperature data quality.
------------
Comments 2:In Section 3.1.2, values that fail the step test are flagged and converted to NaN, and the NaN-marked data are reconstructed using the method described in Section 3.3. However, how is the reliability of the reconstructed data assessed?
Response 2:
Thank you for your valuable comment.
The reliability of the reconstructed data was assessed through the Data Quality Index (DQI), as detailed in Section 3.4. Specifically:
-
After the machine learning (ML) reconstruction using XGBoost, we evaluated both accuracy and completeness of the reconstructed time series.
-
Accuracy was assessed by calculating the Root Mean Square Error (RMSE) between the reconstructed values and the real measurements at the target station (for periods with available data), normalized by the observed temperature range (Equations 5–9).
-
Additionally, a penalty factor was introduced for periods with many missing values to avoid artificially inflating accuracy scores.
-
Completeness was evaluated by comparing the percentage of valid data before and after reconstruction (Equations 3–4).
-
The final DQI combined these two aspects (Equation 10) to provide an overall metric of data quality improvement, allowing an objective assessment of the reliability of the reconstructed series.
As shown in Section 4.2 and Tables 7 and 8, the final DQI values after reconstruction were significantly improved (up to +67% in the worst cases), and RMSE values remained low (0.40°C–0.66°C across stations), supporting the robustness and reliability of the reconstructed data.
------------
Commends 3: In Section 3.1.4, the author defines Thmin and Thmax as the average of the minimum and maximum temperatures from all meteorological stations as the threshold for the entire network. This does not fully account for the local climate characteristics and geographical differences of each station. The climate variations between different stations may cause these average values to be unsuitable for all stations, especially under extreme weather or specific climatic conditions.
Response 3: Thank you for raising this important point. We agree that using a network-wide average for Thmin and Thmax has limitations. To address this, we have added a paragraph in Section 3.1.4 (lines 291 - 301) that explains our rationale for this approach. Specifically, we discuss the spatial homogeneity of the Arta plain, the low altitudes and minimal topographic influence, and the results of our spatial consistency tests (DTW and Spatial Regression), which support the use of network-averaged values. We acknowledge the potential for localized microclimatic effects but believe that the network-averaged values are appropriate for the internal consistency checks performed in this study.
------------
Comments 4:There is confusion in the chapter numbering, and the author needs to renumber the sections.
Response 4:Thank you for pointing out the confusion in chapter numbering. We acknowledge the issue and understand the need for renumbering the sections to improve clarity. However, at this stage of the review process, we have made numerous other changes to the manuscript. We are concerned that reordering the sections might introduce further confusion for the current reviewers, who are already familiar with the existing structure. We would like to address this numbering issue in the revised version, if the paper is accepted, to avoid any potential inconsistencies during this review phase. We are open to suggestions on how to best approach this to minimize disruption.
------------
Comments 5: There are duplicate equation numbers, such as two different equations both labeled as (1).
Response 5: So sorry, equations labeling was corrected.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe paper Enhancing Temperature Data Quality for Agricultural Decision-Making with Emphasis on Evapotranspiration Calculation: A Robust Framework Integrating Dynamic Time Warping, Fuzzy Logic, and Machine Learning addresses a very interesting topic, but it is not optimally structured. In particular, the Materials and Methods section needs to be organized more coherently. Additionally, both the Introduction and Discussion sections are too brief and are not supported by an adequate bibliography. Below are my specific comments:
-
Please review the keywords; it would be preferable to avoid including terms that are already present in the title.
-
The Introduction section should be expanded and supported with more bibliographic references.
-
Check the formatting of Table 1, especially the row referring to the installation year. It would also be advisable to add a closing line to clearly separate the table from the body text.
-
Which climate classification system is used for the climate characterization? Please specify.
-
Section 2 should be reorganized: some parts (e.g., 2.1 and parts of 2.2) would be more appropriately placed under Materials and Methods. The entire description of the errors would be better positioned in the Results section.
-
The Discussion should be expanded, as it is currently too brief and does not adequately analyze the obtained results.
Best regards
Author Response
Comments 1: Please review the keywords; it would be preferable to avoid including terms that are already present in the title.
Response 1: I am considering the keywords for search engine optimization mechanisms that align with the publisher's strategy. However, to ensure they are optimal, I will consult with the publisher for their opinion.
-----------
Comments 2: The Introduction section should be expanded and supported with more bibliographic references.
Respose 2: We thank the reviewer for this insightful comment. In response, we have revised the Introduction section to include a concise review of existing relevant studies. Specifically, we now reference
[4] P. Cerlini, L. Silvestri, and M. Saraceni, “Quality control and gap-filling methods applied to hourly temperature observations over Central Italy,” Meteorol. Appl., vol. 27, 2020.
[5] M. Boujoudar et al., “Comparing machine learning algorithms for imputation of missing time series in meteorological data,” Neural Comput. Appl., Dec. 2024, doi: 10.1007/s00521-024-10601-8.
-----------
Comments 3:Check the formatting of Table 1, especially the row referring to the installation year. It would also be advisable to add a closing line to clearly separate the table from the body text.
Response 3: The row of installation year was removed as is no valuable information, and also measuremnts are start from 2015. Also table 1 separeted from the text body.
-----------
Comments 4: Which climate classification system is used for the climate characterization? Please specify.
Response 4: The following sentense added
"According to the Köppen-Geiger climate classification system (Peel et al., 2007), the climate of the Arta plain is categorized as Csa, indicating a temperate climate with dry, hot summers and mild, wet winters. This classification is consistent with the typical Mediterranean climatic patterns observed in the region."
-----------
Comments 5:Section 2 should be reorganized: some parts (e.g., 2.1 and parts of 2.2) would be more appropriately placed under Materials and Methods. The entire description of the errors would be better positioned in the Results section.
Response 5: At this moments i hesitate to reorganize the paragraphs because i may confuse the other reviewers comments. I will really appreciate to leave it as final step.
Reviewer 4 Report
Comments and Suggestions for Authors- Lines 18–27: The abstract mentions "evapotranspiration calculation" as a research focus, but the results section does not clearly demonstrate how the improved temperature data quality impacts evapotranspiration modeling outcomes.
- Lines 45–47: Some of the cited references are relatively outdated and do not sufficiently reflect recent developments in machine learning-based imputation for meteorological data.
- Lines 93–95: The observation network is classified as "mesoscale" with a station spacing of approximately 7 km. Further clarification is needed to explain the alignment of this classification with WMO standards, as 7 km spacing may be closer to "toposcale" in practice.
- Lines 163–165: The term “contextual outliers” is repeated in two consecutive sentences, which is redundant.
- Lines 274–275: The “internal consistency” test uses the mean of neighboring stations' maximum and minimum temperatures, assuming spatial homogeneity. This assumption needs justification, especially in regions with microclimates.
- Line 343: Equation (1) applies a log transformation to DTW distances with ε = 10⁻⁹, but the impact of this parameter on the results is not discussed.
- Lines 474–477: The selection of XGBoost is justified by its "balanced dependency," but the manuscript lacks quantitative comparisons (e.g., computational efficiency or training time) with other models such as LSTM.
- Line 552: The penalty factor of 0.1 used in Equation (7) is arbitrary. The manuscript does not provide a theoretical or empirical rationale or conduct a sensitivity analysis.
- Line 613: Table 3 presents error counts (e.g., "2542") without specifying whether these are absolute numbers or percentages of the total dataset.
- Lines 623–629: The reported RMSE values (0.40°C–0.66°C) are not compared with baseline methods (e.g., spatial interpolation or simpler ML models), making it difficult to evaluate the superiority of the proposed approach.
- Lines 528–529: Table 7 assumes a "Final Completeness" of 100% after reconstruction. However, machine learning imputation rarely achieves perfect restoration in practice. This limitation is not addressed.
- Lines 709–710: The statement that "at least two or better three weather stations must operate" to ensure data quality is not supported by references or empirical evidence.
- Lines 720–722: While future work mentions extending the framework to other variables (e.g., humidity), it does not discuss scalability challenges such as the computational cost of multi-variable modeling.
- Figures 2 and 5: Figures (e.g., Figure 2 and Figure 5) should be explicitly referenced in the main text, and all axes must include labels and units.
- No specific line – should be added near appendix or acknowledgments: A statement regarding the availability of code and data (e.g., via a GitHub repository) is missing and should be included to enhance reproducibility.
- Line 700: There are minor grammatical errors (e.g., “for from”) that require careful proofreading.
Author Response
Thank you for your valuable feedback.
Comments 1:Lines 18–27: The abstract mentions "evapotranspiration calculation" as a research focus, but the results section does not clearly demonstrate how the improved temperature data quality impacts evapotranspiration modeling outcomes.
Response 1: We thank the reviewer for this insightful comment. To address this, we have added a new Section 4.4 where we apply the Hargreaves–Samani method to estimate reference evapotranspiration (ET₀) using both raw and cleaned temperature data. The results demonstrate that correcting temperature data significantly improves ET₀ during problematic periods. This confirms that the presented temperature quality framework has a direct and meaningful impact on evapotranspiration modeling applications.
----------------------------
Comments 2:Lines 45–47: Some of the cited references are relatively outdated and do not sufficiently reflect recent developments in machine learning-based imputation for meteorological data.
Response 2: New upated references [26], [27], [28] was added
----------------------------
Comments 3: Lines 93–95: The observation network is classified as "mesoscale" with a station spacing of approximately 7 km. Further clarification is needed to explain the alignment of this classification with WMO standards, as 7 km spacing may be closer to "toposcale" in practice.
Response 3: In this work, the agrometeorological station network of the Department of Agriculture of the University of Ioannina (UOI-DAGRI), which serves agricultural applications across the plain of Arta, Greece (Table 1; Figure 1), is classified as a "mesoscale" network (3 km < spatial density < 100 km) according to WMO standards [4]. The station spacing, approximately 7 km, places the network near the boundary between the "toposcale" (100 m–3 km) and "mesoscale" (3–100 km) domains [4,6]. While it formally falls within the mesoscale classification, its relatively dense configuration aligns more closely with the higher-resolution end of mesoscale or even the upper range of toposcale applications.
----------------------------
Comments 4: Lines 163–165: The term “contextual outliers” is repeated in two consecutive sentences, which is redundant.
Response 4: sentence was corrected as
..contextual outliers can arise anomalies that may appear normal when viewed in isolation but are considered outliers when analyzed within their specific temporal or spatial context. In temperature data, these anomalies occur...
----------------------------
Comments 5:Lines 274–275: The “internal consistency” test uses the mean of neighboring stations' maximum and minimum temperatures, assuming spatial homogeneity. This assumption needs justification, especially in regions with microclimates.
Response 5: The following response also added as explanation paragraph
In this study, Thmin and Thmax were calculated as the mean values of the daily maximum and minimum temperatures across all operational stations. This approach relies on the assumption of spatial homogeneity within the network. The assumption is considered reasonable because the agrometeorological stations are installed in the relatively flat and homogeneous landscape of the Arta plain, at low altitudes (0–20 m), with minimal topographic influence. Moreover, spatial consistency tests conducted using Dynamic Time Warping (DTW) and the Spatial Regression Test (Section 4.1) demonstrated high inter-station coherence, with pairwise RMSE values typically below 1°C. These results confirm a strong spatial correlation among stations, supporting the use of network-averaged extreme values (Thmin and Thmax) for internal consistency checks, despite the potential presence of localized microclimatic effects.
----------------------------
Comments 6: Line 343: Equation (1) applies a log transformation to DTW distances with ε = 10⁻⁹, but the impact of this parameter on the results is not discussed.
Response 6: explanation added
... zero-valued distances. This value was selected to be sufficiently small so as not to affect the transformation's outcome for non-zero values. Sensitivity analysis with alternative small values (e.g., 10⁻⁸ and 10⁻¹⁰) showed no notable impact on the distribution characteristics or the resulting anomaly thresholds. Therefore, ε = 10⁻⁹ was adopted as a safe and robust default. The transformation ...
----------------------------
Comments 7:Lines 474–477: The selection of XGBoost is justified by its "balanced dependency," but the manuscript lacks quantitative comparisons (e.g., computational efficiency or training time) with other models such as LSTM.
Response 7: Explanation added
Although no formal benchmark was conducted, XGBoost was selected due to its known advantages over alternatives such as LSTM for the specific needs of this study. These advantages include its native handling of missing values without the need for complex data imputation, its lower computational requirements that make it suitable for CPU-only environments, its faster training and tuning times, which are especially important for near real-time reconstruction in operational networks, and its easier model interpretability, such as the ability to analyze feature importance.
----------------------------
Comments 8: Line 552: The penalty factor of 0.1 used in Equation (7) is arbitrary. The manuscript does not provide a theoretical or empirical rationale or conduct a sensitivity analysis.
Response 8: Explanation paragraph added
The penalty factor of 0.1 was selected as a pragmatic choice to balance penalization without excessively distorting the initial accuracy score. A lower factor (e.g., 0.05) would underestimate the impact of missing data, while a higher factor (e.g., 0.2) would overemphasize it. Although a formal sensitivity analysis was not conducted, preliminary tests varying the penalty factor within the range of 0.05 to 0.2 showed negligible effects on the overall DQI improvement, indicating that the selected value is robust for the intended application.
----------------------------
Comments 9:Line 613: Table 3 presents error counts (e.g., "2542") without specifying whether these are absolute numbers or percentages of the total dataset.
Response 9: Table 3 caption was changed to:
Summary of quality control (QC) test results for each station. The table reports the absolute number of flagged cases, followed in parentheses by the percentage relative to the total dataset (N = 82,967).
----------------------------
Comments 10: Lines 623–629: The reported RMSE values (0.40°C–0.66°C) are not compared with baseline methods (e.g., spatial interpolation or simpler ML models), making it difficult to evaluate the superiority of the proposed approach.
Response 10:
Thank you for your comment. We acknowledge the importance of comparing the RMSE values of our reconstruction approach with baseline methods such as spatial interpolation or simpler machine learning (ML) models. However, it is important to note that such baseline methods are typically not robust to long periods of missing data (common in weather station time series) without first applying separate imputation techniques. For example, traditional spatial interpolation approaches assume continuous data coverage and often cannot function when large gaps exist, as seen in our dataset. Similarly, many simple ML models require complete inputs and do not natively handle NaN values.
One of the key strengths of our approach—specifically the use of XGBoost—is its ability to natively handle missing values and perform reconstruction without the need for prior imputation, even under extended gaps and partial station outages. This capability ensures seamless integration into operational workflows.
We agree that a direct performance comparison with baseline methods is an interesting direction for future work. A follow-up study will be conducted to benchmark the proposed framework against interpolation methods and alternative ML models under identical data loss scenarios.
----------------------------
Comments 11:Lines 528–529: Table 7 assumes a "Final Completeness" of 100% after reconstruction. However, machine learning imputation rarely achieves perfect restoration in practice. This limitation is not addressed.
Response 11: Thank you for your observation. We would like to clarify that in our study, the term "Final Completeness" (Table 7) strictly refers to the absence of missing data after machine learning (ML) reconstruction, i.e., no empty data points remain. It does not imply perfect restoration of the true temperature values. Same for results on table 8. A paragraph to inform reader about terms of "Final Completeness" and "Final Accuracy" was added .
Data Quality Index (DQI) improvement during the Longest Problematic Period for each weather station. "Final Completeness" refers solely to the absence of missing (empty) data after machine learning (ML) reconstruction. "Final Accuracy" represents the accuracy of the reconstructed temperature values, assessed using the RMSE metric normalized by the temperature range.
----------------------------
Comments 12:Lines 709–710: The statement that "at least two or better three weather stations must operate" to ensure data quality is not supported by references or empirical evidence.
Response 12: The proposed DTW-Fuzzy Logic test relies on cross-station comparisons to identify contextual outliers. Therefore, maintaining at least two, and preferably three, operational weather stations in the vicinity is critical to ensure the robustness of anomaly detection, as more comparisons improve reliability and reduce the likelihood of undetected errors. The follows papragraph was added to support it.
"This recommendation is directly supported by the operational principles of the DTW-Fuzzy Logic anomaly detection framework proposed in this study. Since the DTW algorithm assesses the similarity between temperature time series from different stations, and the Fuzzy Logic model refines anomaly detection based on multiple cross-comparisons, having multiple stations operational simultaneously significantly strengthens the robustness and reliability of outlier identification. When only a single neighboring station is available, the capacity to distinguish between true anomalies and localized meteorological variations is reduced. Therefore, the simultaneous operation of at least two, and ideally three, weather stations within a mesoscale network is critical to maintain high spatial redundancy, enhance anomaly detection sensitivity, and safeguard overall data quality."
----------------------------
Comments 13: Lines 720–722: While future work mentions extending the framework to other variables (e.g., humidity), it does not discuss scalability challenges such as the computational cost of multi-variable modeling.
Response 13: The paragraph was improved.
"Planning for future research includes the application of the proposed approach for other parameters such as relative humidity and rain and its combination with remote sensing data for the creation of virtual agro-meteorological stations. Initially, the extension of the framework to other variables (e.g., humidity) will be carried out independently; nonetheless, future work will also investigate multivariate modeling strategies to assess potential benefits for anomaly detection and reconstruction. Scalability challenges such as the computational cost of multi-variable modeling are expected to be manageable, as both the Dynamic Time Warping (DTW) method and the XGBoost model demonstrated fast computational performance during this study."
----------------------------
Comments 14: Figures 2 and 5: Figures (e.g., Figure 2 and Figure 5) should be explicitly referenced in the main text, and all axes must include labels and units.
Response 14: Figures references was made
----------------------------
Comments 15: No specific line – should be added near appendix or acknowledgments: A statement regarding the availability of code and data (e.g., via a GitHub repository) is missing and should be included to enhance reproducibility.
Response 15:
Thank you for your valuable suggestion. We agree that data availability enhances reproducibility. While we have chosen not to make the code publicly available at this time, we are happy to share the input datasets used in this study. We have added the following statement near the Appendix section:
Data Availability: The input datasets used in this study are available from the corresponding author upon reasonable request. The results of the data processing and quality control procedures for each day of the examined timeframe of this work can be accessed at: https://ckoliopanos.github.io/Cleared_temperature_data/.
----------------------------
Comments 16: Line 700: There are minor grammatical errors (e.g., “for from”) that require careful proofreading.
Response 16: Grammatical error corrections as:
This study presents a data quality assessment framework for hourly temperature data collected from a six-station agrometeorological network...
----------------------------
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors have responded and revised each comment thoroughly, but there are still some minor errors: for example, some of the tables in the article have incomplete borders, and the Discussion and Conclusions sections have incorrect serial numbers.
Author Response
Thank you for your comments.
Comment 1: The authors have responded and revised each comment thoroughly, but there are still some minor errors:
1. For example, some of the tables in the article have incomplete borders, and the
2. Discussion and Conclusions sections have incorrect serial numbers.
Response 1:
1. You are probably referring to Table 2. The formatting and readability of Table 2 have been improved. All other tables have also been checked to ensure they conform to the Journal's template.
2. The numbering of the Discussion and Conclusions sections has been corrected.
Reviewer 2 Report
Comments and Suggestions for AuthorsI am glad that the author has addressed the issues I am concerned about accurately and in detail. I have no further suggestions. Therefore, I would like to recommend the manuscript for publication.
Author Response
Comments: I am glad that the author has addressed the issues I am concerned about accurately and in detail. I have no further suggestions. Therefore, I would like to recommend the manuscript for publication.
Response: Thank you very much for accepting my paper for publication. I sincerely appreciate the time and effort you dedicated to reviewing my work.
Reviewer 3 Report
Comments and Suggestions for AuthorsDear authors,
What is the editor's opinion about keywords? It would be preferable to avoid including terms that are already present in the title.
Regarding the tables, it would be better to revise also table 2.
Before approving the article, it would be better if the paragraphs were reorganized as recommended during the first round of review: Section 2 should be reorganized: some parts (e.g., 2.1 and parts of 2.2) would be more appropriately placed under Materials and Methods. The entire description of the errors would be better positioned in the Results section.
Best regards
Author Response
Comment1:
1.1 What is the editor's opinion about keywords? It would be preferable to avoid including terms that are already present in the title.
1.2 Regarding the tables, it would be better to revise also table 2.
1.3 Before approving the article, it would be better if the paragraphs were reorganized as recommended during the first round of review: Section 2 should be reorganized: some parts (e.g., 2.1 and parts of 2.2) would be more appropriately placed under Materials and Methods. The entire description of the errors would be better positioned in the Results section.
Response 1:
1.1
We appreciate the editor’s suggestion concerning the selection of keywords. Editor asked and his opinion aligns with yours. We fully agree that it is preferable to avoid repeating terms already present in the title. Accordingly, the keyword list has been revised to focus on specific techniques and applications relevant to the study, while avoiding redundancy with the title. The updated keywords are:
“Anomaly detection; Dynamic Time Warping (DTW); Fuzzy Logic; XGBoost; Spatiotemporal analysis; Irrigation scheduling; Climate data cleaning; Evapotranspiration modeling”
1.2
Thank you for pointing out the need to revise Table 2. The table has now been reformatted to improve clarity and readability. Each test and its corresponding formulation are clearly presented, ensuring that the information is accessible and easy to interpret for readers.
1.3
We sincerely thank the reviewer for this helpful comment regarding the structure of our manuscript. In response, we have made the following changes:
-
Section 2.1 (“Area Description and Weather Station Network”) has been relocated to Section 3 (“Materials and Methods”), as it describes the infrastructure and operational setup of the monitoring network, which supports the methodological framework of the study.
Also the following sentense "In addition to the data-specific challenges described above, maintaining the continuous operation of an agrometeorological network presents further technical and resource-based difficulties." added to the beggining of 2.2 to help guide the reader logically move from data problems to operational/systemic challenges. -
Section 2 has been retitled to “Temperature Dataset Characteristics and Data Quality Challenges” to better reflect its new focus on the nature of the temperature data and the issues encountered.
- Paragraphs numbering was reorganized.
We have, however, retained the discussion of observed error types and data-related challenges within Section 2 rather than moving it to the Results section. This decision was made for the following reasons:
-
Establishing Problem Context:
Presenting the types of observed data errors early in the manuscript provides essential context for understanding the necessity and design of the data quality control and reconstruction methods introduced in Section 3. -
Maintaining Narrative Logic:
Describing data issues before the methodology allows for a clearer, problem-driven flow—moving from identifying real-world measurement challenges to proposing specific technical solutions. Placing these descriptions in the Results section would disconnect the motivation from the method. -
Consistency with Scientific Practice:
In environmental monitoring and quality control studies, it is common practice to present the nature of the dataset and its inherent limitations prior to describing the applied methodologies. This helps readers follow the rationale behind the selection of techniques.
We hope this revised organization addresses the reviewer’s concerns while preserving a coherent and logical structure. Of course, we remain open to further editorial input should additional restructuring be preferred.
Reviewer 4 Report
Comments and Suggestions for AuthorsIt can be accepted in present form.
Author Response
Comments: It can be accepted in present form.
Response: Thank you very much for accepting my paper for publication. I sincerely appreciate the time and effort you dedicated to reviewing my work