An Efficient Downwelling Light Sensor Data Correction Model for UAV Multi-Spectral Image DOM Generation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper tackles a relevant and practical issue in UAV-based multispectral remote sensing—errors in Downwelling Light Sensor (DLS) data caused by flight dynamics during rotary-wing UAV operations. The authors introduce a post-processing correction model (FIM-DC) aimed at improving reflectance accuracy under clear-sky conditions. The topic is timely and important for applications such as precision agriculture and environmental monitoring. Below are my questions and comments:

 

1. The abstract states that the FIM-DC model "further refines DLS measurements under stable illumination conditions" and is "specifically tailored for DLS data collected by rotary-wing UAVs under clear-sky conditions." However, the introduction emphasizes that DLS measurements can become unreliable due to external factors such as "strong wind gusts" and "significant flight attitude anomalies. It should be clarified whether these conditions are strict requirements or just optimal scenarios. A brief discussion on the method’s performance under moderate illumination variability (e.g., partial cloud cover) would enhance understanding of its practical limitations.
2. Why are only clear-sky conditions considered? Have the authors attempted any validation under slightly variable atmospheric conditions?
3. The equation formatting (Equations 1–3) and Figure 1 are not very clear. Could you clarify or improve the math presentation and diagrams if possible?
4. Why is the ground ASD spectroradiometer only used in the Jiading experiment and not Qingpu?
5. What is the rationale for selecting a quadratic fitting curve in the FIM-DC model? Have alternative fitting approaches (e.g., linear, cubic, or spline-based models) been considered? If so, how does the quadratic model perform in terms of accuracy, computational efficiency, and robustness across varying flight conditions?
6. Have the authors considered using the UAV’s pitch, roll, and yaw data directly in the correction? This might help improve accuracy during quick movements, rather than relying only on irradiance curve fitting.
7. There are a few grammar inconsistencies—for example, switching between phrases like "further refines DLS measurements" and "corrects DLS data."
8. References are comprehensive, but if i am not wrong, some entries (e.g., duplicate citation of Swaminathan et al., 2024) should be consolidated.

Author Response

• The abstract states that the FIM-DC model "further refines DLS measurements under stable illumination conditions" and is "specifically tailored for DLS data collected by rotary-wing UAVs under clear-sky conditions." However, the introduction emphasizes that DLS measurements can become unreliable due to external factors such as "strong wind gusts" and "significant flight attitude anomalies. It should be clarified whether these conditions are strict requirements or just optimal scenarios. A brief discussion on the method’s performance under moderate illumination variability (e.g., partial cloud cover) would enhance understanding of its practical limitations.

Thank you very much for this insightful comment. We agree that this distinction is critical for understanding the scope of our model. In response, we have explicitly clarified in both the abstract and the revised manuscript that clear-sky and cloud-free conditions are indeed strict prerequisites for the application of the FIM-DC model(line 13-17). Additionally, we have included a forward-looking statement(line 421-431) emphasizing that future studies will aim to evaluate the model’s performance under moderate illumination variability, such as during partially cloudy weather. These updates are now reflected in the revised abstract and Section 1 (Introduction), with the relevant changes highlighted.

• Why are only clear-sky conditions considered? Have the authors attempted any validation under slightly variable atmospheric conditions?

We sincerely appreciate your attention to the applicability of our method under varying atmospheric conditions. As mentioned above, the current implementation of the FIM-DC model is specifically designed and validated only under clear-sky, cloud-free conditions, which ensures stable irradiance input for the DLS system. At this stage, we have not conducted validation under more complex weather scenarios. However, as you rightly suggested, we recognize the importance of extending this work to more variable atmospheric conditions, such as partially cloudy or hazy skies, and have added this direction as a focus for future work in the Conclusion section.

• The equation formatting (Equations 1–3) and Figure 1 are not very clear. Could you clarify or improve the math presentation and diagrams if possible?

Thank you for pointing this out. We have carefully revised the formatting of Equations (1) to (3) to improve mathematical clarity and consistency (line144-153). Additionally, we have updated Figure 1 to enhance its visual quality and layout for better readability. These improvements have been incorporated into the revised manuscript, and all corresponding changes are highlighted to facilitate your review.

• Why is the ground ASD spectroradiometer only used in the Jiading experiment and not Qingpu?

We appreciate your thoughtful question. Due to equipment scheduling conflicts and institutional ownership constraints, we were unfortunately unable to access the ASD spectroradiometer during the Qingpu experiment. As a result, the device was used solely in the Jiading experiment. We regret this limitation and have added an explanation to clarify this in the revised manuscript (Section 3.1). Thank you for your understanding.

• What is the rationale for selecting a quadratic fitting curve in the FIM-DC model? Have alternative fitting approaches (e.g., linear, cubic, or spline-based models) been considered? If so, how does the quadratic model perform in terms of accuracy, computational efficiency, and robustness across varying flight conditions?

Thank you for raising this important technical point. The decision to employ a quadratic curve fitting in the FIM-DC model is based on the operational characteristics of rotary-wing UAVs. Typically, flight missions span about 20–30 minutes, during which the DLS data exhibit a stable and smooth trend, often either gradually increasing or decreasing. As such, a quadratic function is sufficient to model this behavior accurately. More complex patterns, such as inflection points or trend reversals within a single flight, are uncommon. We have elaborated on this rationale in Section 2.3 of the revised manuscript.

• Have the authors considered using the UAV’s pitch, roll, and yaw data directly in the correction? This might help improve accuracy during quick movements, rather than relying only on irradiance curve fitting.

We truly appreciate your constructive suggestion. In fact, the FIM-DC model is fundamentally built upon UAV attitude correction. We have added a more detailed explanation in Section 2.2 (line 157-166) to describe our image preprocessing workflow. Specifically, before irradiance fitting and filtering, we utilize UAV pitch, roll, and yaw data to perform attitude correction on the DLS measurements, generating horizontal irradiance values. The subsequent steps, including curve fitting and outlier removal, are performed based on these corrected values. Thank you again for highlighting this point.

• There are a few grammar inconsistencies—for example, switching between phrases like "further refines DLS measurements" and "corrects DLS data."

Thank you for your careful review of the manuscript's language. We sincerely apologize for the inconsistency in terminology. In the revised version, we have carefully reviewed and standardized the language used throughout the manuscript to ensure consistency and clarity, especially regarding expressions such as “refinement” and “correction” of DLS data. We appreciate your attention to detail.

• References are comprehensive, but if i am not wrong, some entries (e.g., duplicate citation of Swaminathan et al., 2024) should be consolidated.

Many thanks for this valuable observation. We apologize for the oversight. We have carefully reviewed and corrected all reference entries, ensuring that any duplicate citations (including that of Swaminathan et al., 2024) have been properly consolidated. The revised bibliography now reflects these changes.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The article addresses a critical challenge in UAV-based multispectral imaging—accurate downwelling light sensor (DLS) data correction for digital orthophoto map (DOM) generation. The authors clearly identify the limitations of existing correction methods, particularly the reliance on angle compensation, which struggles with timing mismatches and UAV flight instabilities such as sudden turns or wind gusts. This clear problem statement sets a strong foundation for the study. The introduction of the FIM-DC (Fitting and Interpolation Model-based Data Correction) method builds logically on existing work and targets real-world issues that degrade image quality. The study’s results, supported by empirical data from the Qingpu region, demonstrate a significant reduction in reflectance variability, which is an impressive improvement (from 15.07% to 0.58% standard deviation). Additionally, the method’s ability to correct spectral curves across multiple land cover types highlights its practical applicability and robustness. The abstract concisely presents these findings and emphasizes the enhancement in reflectance accuracy and overall image quality, making a strong case for the utility of the proposed method.

While the study presents promising results, it could benefit from more clarity regarding the technical details of the FIM-DC model. Specifically, a brief explanation of the core mechanism—how fitting and interpolation are applied to correct the DLS data—would help readers better grasp the innovation’s novelty. The mention of “stable illumination conditions” raises questions about how the method performs under varying or unstable lighting, which are common in many UAV missions; addressing this limitation explicitly would strengthen the scope of the study. Furthermore, although the Qingpu region results are promising, the generalizability to other environments or UAV platforms is not discussed, leaving readers uncertain about the method’s wider applicability. Lastly, the abstract could be improved stylistically by reducing minor redundancies and tightening phrasing for easier readability, helping the key contributions stand out more clearly.

Author Response

• Specifically, a brief explanation of the core mechanism—how fitting and interpolation are applied to correct the DLS data—would help readers better grasp the innovation’s novelty. The mention of “stable illumination conditions” raises questions about how the method performs under varying or unstable lighting, which are common in many UAV missions; addressing this limitation explicitly would strengthen the scope of the study. Furthermore, although the Qingpu region results are promising, the generalizability to other environments or UAV platforms is not discussed, leaving readers uncertain about the method’s wider applicability. Lastly, the abstract could be improved stylistically by reducing minor redundancies and tightening phrasing for easier readability, helping the key contributions stand out more clearly.

Thank you very much for your insightful and constructive suggestions.Firstly, regarding your comment on providing a brief explanation of the core mechanism of the method, we have revised Section 2.1 (line 145-153) to clarify how curve fitting and interpolation are applied for DLS data correction. We have also highlighted the relevant modifications to facilitate reader comprehension.Secondly, concerning your point about the "stable illumination conditions," we have explicitly stated in both the abstract and the main body of the manuscript that clear-sky and cloud-free conditions are strict prerequisites for applying the FIM-DC model. We also plan to explore its performance under moderate illumination variability (e.g., partially cloudy skies) in future studies.As for the generalizability of the model across different UAV platforms, we believe that the FIM-DC model is well-suited for all rotary-wing UAV platforms, as it effectively addresses attitude instability issues commonly observed during UAV turns—particularly in such aircraft types.Finally, we have carefully revised the abstract to eliminate minor redundancies and improve phrasing for enhanced clarity and readability.Once again, thank you for your thoughtful review and helpful feedback.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript proposes a data correction method based on fitting and interpolation models (FIM-DC) to enhance the accuracy of DLS data in UAV multispectral image DOM generation under clear skies. It has practical and promotional value. But there are some revisions and explanations to address to improve readability and presentation as follows:

1）The pre- processing of UAV images before DLS data correction isn't clearly explained in the Experimental section.

2）The article doesn't sufficiently discuss how to effectively determine the upper and lower limits of reflectance for assessing DLS data normality during calibration.

3）It's unclear whether the article directly replaces outliers with two valid data points. Also, the handling of consecutive outliers isn't explained.

4）The method is only tested in Qingpu and Jiading areas with similar geography and climate. More diverse terrain and weather conditions should be included for a comprehensive evaluation of FIM - DC's effectiveness and robustness.

5）As the study focuses on a single - day weather scenario, the method's correction effects on different UAV types should be further investigated.

6）The article mentions IMU data for attitude correction but lacks in - depth analysis linking it to DLS data. A more detailed analysis of the UAV's attitude changes (e.g., roll, pitch, yaw angles) and DLS data anomalies is suggested.

7）While the method is simple and practical, the article could benefit from a discussion on its future prospects.

8）Figure 1's explanation in the text is too vague. The specific meanings and functions of the diagram's components should be clarified.

9）Figures 6 and 7 lack detailed in - text explanations and analysis. Their labels are also unclear, such as not specifying image names and reflectance values.

Author Response

• The pre- processing of UAV images before DLS data correction isn't clearly explained in the Experimental section.

Thank you very much for your insightful suggestion. In response, we have revised Figure 2. Experimental Flowchart and the corresponding textual description in Section 2.2 Experimental Procedure (line 157-166). We have now clearly detailed the UAV image pre-processing steps, including Radiometric Correction (Dark Current & Flat-Field Correction, Irradiance Normalization) and Geometric Correction (Image Registration, Orthorectification, and Band Co-registration), which are conducted prior to DLS data correction.

• The article doesn't sufficiently discuss how to effectively determine the upper and lower limits of reflectance for assessing DLS data normality during calibration.

We sincerely appreciate your valuable comment. We have now addressed this point in Section 5.3 (line 379-381) Analysis of Direct and Scattered Radiation Proportions in DLS Data, where we provide a detailed explanation of how the upper and lower reflectance thresholds are determined. Furthermore, we discuss the influence of these thresholds on filtering performance and reflectance calculation accuracy.

• It's unclear whether the article directly replaces outliers with two valid data points. Also, the handling of consecutive outliers isn't explained.

Thank you for raising this important concern. We acknowledge that the original manuscript lacked clarity in this aspect. Therefore, we have revised Figure 1 and expanded the explanation in Section 2.1 Construction of FIM-DC (line 145-154) to clearly describe how outliers, including consecutive ones, are detected and corrected using linear interpolation between adjacent valid data points. This clarification ensures that the mechanism is now fully transparent to readers.

• The method is only tested in Qingpu and Jiading areas with similar geography and climate. More diverse terrain and weather conditions should be included for a comprehensive evaluation of FIM - DC's effectiveness and robustness.

Thank you for this valuable recommendation. We would like to clarify that the FIM-DC model is explicitly designed for application under clear-sky, cloud-free conditions, as stated in both the Abstract and main text. Nonetheless, we acknowledge the need for broader testing, and we have highlighted in the line 421-431 that future research will explore the model’s applicability under moderate illumination variability (e.g., partially cloudy conditions).

• As the study focuses on a single - day weather scenario, the method's correction effects on different UAV types should be further investigated.

We appreciate your constructive feedback. The FIM-DC model is designed to be applicable across all rotary-wing UAV platforms. Its robustness stems from its ability to address radiometric inconsistencies caused by attitude instability, which is a common issue across various rotary-wing UAVs during flight turns. We have clarified this point in the revised manuscript.

• The article mentions IMU data for attitude correction but lacks in - depth analysis linking it to DLS data. A more detailed analysis of the UAV's attitude changes (e.g., roll, pitch, yaw angles) and DLS data anomalies is suggested.

Thank you for this important observation. The FIM-DC model is fundamentally built upon UAV attitude compensation. We have expanded Section 2.2 (line 157-166) to explain how roll, pitch, and yaw data obtained from the IMU are used to correct the DLS-recorded irradiance values. This correction serves as the foundation upon which the subsequent data filtering and interpolation steps of the FIM-DC model are applied.

• While the method is simple and practical, the article could benefit from a discussion on its future prospects.

We are grateful for your suggestion. In response, we have expanded the Conclusion section (line 421-431) to discuss the future development of the FIM-DC model, including its potential integration with real-time onboard correction frameworks and its performance validation under diverse environmental conditions. We also acknowledge current limitations and propose future research directions.

• Figure 1's explanation in the text is too vague. The specific meanings and functions of the diagram's components should be clarified.

Thank you for highlighting this issue. We have revised the formatting and description of Equations (1)–(3) and Figure 1 (line144-153), with improved mathematical presentation and clearer annotations. The updated diagram now includes detailed labels and descriptions of each component to enhance interpretability, and we have highlighted these changes in the revised manuscript.

• Figures 6 and 7 lack detailed in - text explanations and analysis. Their labels are also unclear, such as not specifying image names and reflectance values.

We truly appreciate your detailed review. Figures 6 and 7 are respectively linked to Tables 3 and 4. In the revised manuscript, we have improved the clarity of both figures and tables, updated the figure legends, and expanded the corresponding text to provide explicit image names, reflectance values, and their comparative interpretations. These revisions improve readability and support clearer understanding of the results.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript proposes a data correction method based on fitting and interpolation models (FIM-DC) to enhance the accuracy of DLS data in UAV multispectral image DOM generation under clear skies. The revisions and suggestions previously mentioned have been modified in the revised manuscript. There are no other issues.I suggest that this article could be considered for acceptance.