Identification of Pine Wilt Disease-Infested Stands Based on Single- and Multi-Temporal Medium-Resolution Satellite Data

Round 1

Reviewer 1 Report

The authors of the article solve a very important problem in the field of control and prevention of pine forests against the harmful disease PWD (cancer of the pine tree). To improve the accuracy and efficiency of monitoring of this tree disease, the authors constructed a model  monitoring forests (infested by PWD) by combining multi-temporal and single-time-phase Landsat remote sensing data, based on application of three machine learning algorithms (ANN, FR and SVM).

The paper is written in a systematic manner. Personally, I enjoy the content of the paper very much. Paper presents solving strategy based on the evaluation of the satellite  remote sensing data (Table 1). The authors visually interpreted and distinguished healthy and infested wood, using the presence or absence of discolored pine in the sample plots as criteria.

The issues raised in the work are up to date. The working methodology is correct. However, several elements require clarification before publication. Hovewer, I kindly ask the authors to consider the following:

1. Specify the architecture of the artificial neural network, including the layers and their hyperparameters. This would improve the quality of work.

2. Provide a more detailed description of (2a) how you used the mentioned algorithms, (2b) how you selected optimization settings when creating models, and (2c) which settings proved to be the best. The specific description ensures comparability with other works.

3. Please, summarize the results of ANOVA in a table. I tis important for the benefit of the reader.

4. Every equation numbered must be cited, otherwise remove the number of the equation.

5. Figure 2 – please correct in caption of Figure 2  „... data captured September 13, the year is missing here, (b) Partial image of the UVA remote sensing data captured October 1, 2022.

6. Please, consider to perform another proof reading, because there are some mistakes, e.g. Figures in the text must be cited as “Figure 1” (instead of Fig. 1). In Section 2.1., there is (Fig. 1); Fig. 2 in Section 2.2.1.;  Fig. 3 in Section 3.1., fig. 4 in Section 3.1. In Section 3.3. there is “the latter by 8.30% (Figures 7) -> correct it to Figure 7. To There are missing space before cited references: “…discolored pine trees[15]. Consequently, ..“  The same in  Section 2.4.1.: vegetation[22] “research[14,20], in Section 2.5. trees[38]. etc.  Please check it  in the manuscript.

I believe that if the author follow these recommendations, the paper will be ready for publication.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper submitted to me for review entitled "Identification of Pine Wilt Disease infested stands based on single and multi-temporal medium resolution satellite data" is a very interesting study. After reading the manuscript, I have a few comments that I believe will allow a wider audience to better understand the results presented.

Abstract:

In the abstract, clarify the main focus and contribution of the study to provide the reader with a concise summary.

Highlight the importance of the results in terms of their implications for PWD monitoring and management and emphasise the potential impact on forest health and conservation efforts.

1.            Introduction:

Begin the introduction with a definition of pine wilt disease (PWD) and its ecological impacts, including its effects on pine stands and forest ecosystems.

Justify the choice of Landsat-9 and Sentinel-2 satellites for pine wilt detection and emphasise their spatial coverage, spectral resolution and historical data availability.

Clearly state the objectives of the study, including the comparison of Landsat-9 and Sentinel-2 data for PWD identification and the exploration of multi-temporal Landsat data to improve monitoring accuracy.

2. Materials and methods

2.1. Overview of the study area:

Provide a brief rationale for the selection of Muping District, Yantai City, Shandong Province, China, as the study area and highlight its significance in terms of the severity of pine wilt disease (PWD) infestation and ecological impacts on local forest ecosystems.

Clarify the criteria used to select the five pine forests as experimental sites and emphasise the representativeness of these sites in terms of the occurrence of PWD and the dominance of Japanese black pine (Pinus thunbergii Parl.).

Consider including additional details about the spatial extent and topographic features of the study area to provide readers with a better understanding of the ecological context in which the research was conducted.

2.2. Data Acquisition:

2.2.1 Remote sensing data from UAV:

Provide the specific objectives of collecting UAV visible light (RGB) imagery three to four times per month and emphasise its relevance to the temporal dynamics of PWD infestation and symptom expression in pine trees.

Provide details of the flight parameters and image acquisition protocols used during UAV data collection, including the choice of flight altitude and timing of image acquisition in relation to the phenological stages of pine health and disease progression.

2.2.2 Satellite remote sensing data:

Provide the specific spectral bands and sensor characteristics used for the Landsat series satellites (Landsat 5, 7, 8, and 9) and the Sentinel-2 satellite to provide the reader with a comprehensive understanding of the remote sensing data sources used in the study.

2.3 Data processing:

2.3.1 Satellite data processing:

Provide a more detailed explanation of the image section, radiometric calibration, and atmospheric correction procedures used for the Landsat series satellites, including specific software tools or algorithms used to improve the accuracy and reliability of the satellite imagery.

2.3.2 UAV data processing:

Clarify the process of geographic alignment of UAV data with satellite data to ensure spatial consistency and accuracy when analysing healthy and infested forest plots.

3. Results

3.2. Effects of satellite remote sensing data resolution on model accuracy:

Explain the impact of resampling Sentinel-2 data to achieve higher spatial resolution, including any trade-offs or limitations associated with this approach in terms of data quality, processing time and computational resources.

Provide a more detailed interpretation of the results of the one-way ANOVA comparing model accuracy between Sentinel-2 data at 10 m and 30 m resolution and discuss the statistical significance of the observed differences and possible implications for practical applications of the monitoring models.

4. Discussions:

Expand the discussion on the implications of the study results for understanding the ecological dynamics of PWD infestation in forest ecosystems, including potential implications for forest management practises and strategies to mitigate the spread of PWD.

Discuss the overall relevance of the study results for the further development of remote sensing techniques in forest health monitoring and pestand disease control and highlight innovative methods or findings that could influence future research in this area.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have followed all my comments, for which I am very grateful. In my opinion, the paper in its present form is suitable for publication in the journal Forests