Global-PCG-10: a 10-m global map of plastic-covered greenhouses derived from Sentinel-2 in 2020

Niu, Bowen; Feng, Quanlong; Qiu, Bingwen; Su, Shuai; Zhang, Xinmin; Cui, Rongji; Zhang, Xinhong; Sun, Fanli; Yan, Wenhui; Zhao, Siyuan; Shi, Hanyu; Ou, Cong; Yan, Xiaolu; Gong, Jianhua; Yin, Gaofei; Huang, Jianxi; Liu, Jiantao; Gao, Bingbo; Yao, Xiaochuang; Yang, Jianyu; Zhu, Dehai

doi:https://doi.org/10.5194/essd-2024-538

Preprints

https://doi.org/10.5194/essd-2024-538

Preprints

09 Jan 2025

| 09 Jan 2025

Status: a revised version of this preprint was accepted for the journal ESSD and is expected to appear here in due course.

Global-PCG-10: a 10-m global map of plastic-covered greenhouses derived from Sentinel-2 in 2020

Bowen Niu, Quanlong Feng, Bingwen Qiu, Shuai Su, Xinmin Zhang, Rongji Cui, Xinhong Zhang, Fanli Sun, Wenhui Yan, Siyuan Zhao, Hanyu Shi, Cong Ou, Xiaolu Yan, Jianhua Gong, Gaofei Yin, Jianxi Huang, Jiantao Liu, Bingbo Gao, Xiaochuang Yao, Jianyu Yang, and Dehai Zhu

Abstract. Plastic-covered greenhouse (PCG) is widely used in agricultural production due to its temperature control, water conservation, and wind protection characteristics, significantly enhancing crop yields and economic benefits. However, its long-term and extensive use can lead to environmental issues, such as the accumulation of local toxic gases and the degradation of soil physicochemical properties. Therefore, obtaining a comprehensive distribution of PCGs is essential. To monitor PCGs on a large scale, this study developed a novel approach for producing the first global 10-meters PCGs dataset (Global-PCG-10) with high-quality. Firstly, the globe was divided into multiple 5-degree grids, and grids for classification were organized based on global cropland layer. Then, multi-temporal Sentinel-2 data and initial labels of PCGs were obtained through Google Earth Engine (GEE) to create a training set for deep learning. Next, initial labels were optimized with the active learning strategy combined with the deep learning model, APC-Net. Finally, the PCGs classification results were predicted, spatially analyzed, and compared with publicly released land use and land cover (LULC) datasets. Experimental results indicate that the proposed Global-PCG-10 dataset has a high overall accuracy of 92.08 %. The global area of PCGs is 14,259.85 km², and 69.24 % of PCGs are located in Asia, covering around 9,874.51 km². China has the largest PCGs area of 8,224.90 km², accounting for 57.67 % of the globe and 83.29 % of Asia. Comparisons with other LULC datasets revealed that PCGs, which should be classified as cropland, are often misclassified as bareland, impervious surfaces, ice/snow, etc.

Received: 14 Nov 2024 – Discussion started: 09 Jan 2025

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 9022 KB)

Supplement (434 KB)

Download & links

Status: closed

RC1:
'Comment on essd-2024-538', Anonymous Referee #1, 16 Jan 2025

Global-PCG-10: a 10-m global map of plastic-covered greenhouses derived from Sentinel-2 in 2020
Manuscript number: essd-2024-538
General comments.
This very interesting paper use machine learning and deep learning on Sentinel-2 10 m GSD images to obtain a global map of plastic-covered greenhouses (PCGs). Really, it is not the first global PCG maps since Tong et al. (2024) already published other global PCG map derived from PlanetScope images, a commercial satellite with 3 m GSD, but using also in the first steps Sentinel-2. Although both works have a similar objective (i.e., to attain a global PCG map), the strategies used were quite different.
Thanks to the attained global PCG map, Niu et al. (2025) give interesting data about the area of PCG around the world, the major concentrations, spatial distribution, etc.
The manuscript is well written and it is worth being published. However, a few specifics comments should be taken into account.
Specific comments.
- Some cites in the manuscript appears with an extra comma. For example, in Page 2, Line 64, the cites “Aguilar et al., (2016) and Yang et al., (2017) independently developed…” should be “Aguilar et al. (2016) and Yang et al. (2017) independently developed…” Similarly, “Zhang et al., (2022a)” in Page 2, Line 66, should be “Zhang et al. (2022a)”. Please, correct this issue throughout the manuscript.
- Page 3, Line 71. The cite (Zhang et al., 2024) should be (Zhang et al., 2024a).
- Page 3, Line 76. Zhang et al., 2023 should be 2023a or 2023b. Please, review it.
- Page 5, Line 139-140. “Actually, Sentinel-2 is a constellation consisting of two satellites, i.e., Sentinel-2A and Sentinel-2B, which are in the same sun-synchronous orbit while phased at 180° to each other”. In fact, there is a new Sentinel-2C. You should speak a little about it.
- Page 6, Line 163. In Figure 2 (Stage 2) the train/validation ratio is 7:3, and in the manuscript you wrote 8:2. Is it a mistake in the Figure?
- Page 7, Line 175. In the caption of Figure 3, you should clarify that the size of the reference samples (512×512) are pixels and not meters.
- Page 9, Line 200. In the caption of Figure 4, it is written Multiple-temporal NDVI. Is not more appropriated multi-temporal NDVI?
- Page 9, Line 205. “1> Spectral features”. Strange login method.
- Page 10, Line 212. “2> Textural features”. Strange login method.
- Page 11, Line 237. You should clarify also in the manuscript that the size of the reference samples (512×512) are pixels and not meters.
- Page 12, Line 273. In Figure 2 (Stage 2) the train/validation ratio is 7:3, and in the manuscript you wrote 8:2. Please review it.
- Page 13, Line 286. Fu et al. (2021) is not in reference section.
- Page 15, Line 332-335. There are some numbers without thousands separation (e.g., 9874.51 km2, 2530.56 km2, 8224.90 km2).
- Page 16, Line 344. Figure 8a is not cited in the manuscript, and it should be.
- Page 18, Line 375. Why 20500 points for GH and 20500 for Non-GH. Justify this figure.
- Page 19, Line 381. Table 1 shows the confusion matrix where OA, User Accuracy (UA) and Producer Accuracy (PA) are depicted. Really, UA=Recall and PA=Precision, so, Table 2 is not necessary. The only data useful in Table 2 is F1 Score. I think that you should rewrite the methods and results about the accuracy assessment. Furthermore, Why is the classification so biased? For example, UA is 99.99% and PA is 86.30% for Non-GH and, UA is 84.18% and PA is 99.99% for GH.
- Page 22, Line 421-422. “… and in May 2024, the University of Copenhagen published a global 3-m PCGs dataset also in 2019”. Please, you should cite Tong et al. (2024) here.
- Page 22, Line 430. You should cite Tong et al. (2024) properly in the caption of Figure 12.
- Page 23, Line 434. “Tong et al., (2024) acquired from 3-m …”. Again, this cite appears with an extra comma.
- Page 30, Line 667. “Zhang, X., Liu, L., and Chen, X.: Global annual wetland dataset Data Descriptor at 30 m with a fine classification system from 2000 to 2022, Sci. Data, https://doi.org/10.1038/s41597-024-03143-0, 2024c”. This reference do not appear in the manuscript.
Final Comments:
It is very important that the global PCG map and the code are accessible to researchers. I have tested that the code for generating the initial labels of PCGs is publicly available via the following link on Google Earth Engine: https://github.com/MrSuperNiu/Greenhouse_Classification_GEE. It consists of feature extraction, RF classification, etc. Additionally, the code of APC-Net is accessible through the following link: https://github.com/MrSuperNiu/APCNet. The Global-PCG-10 dataset is stored on figshare, and can be downloaded here: https://doi.org/10.6084/m9.figshare.27731148.v2 (Niu et al., 2024).

Citation: https://doi.org/10.5194/essd-2024-538-RC1
- AC1: 'Reply on RC1', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC1
CC1:
'Comment on essd-2024-538', Liang He, 06 Feb 2025

This article used deep learning methods to produce a refined distribution of global Plastic-covered greenhouse in 2020. The methodology of the paper is sound and the data is novel. Although there are some global PCG data in the current study, the data set has some value in terms of resolution. There is some value for global agricultural macro-management.

There are some problems and deficiencies that need to be corrected as follows:
1. Plastic-covered greenhouse can divided into daylight greenhouses and plastic greenhouses. Is there a significant difference in the spectra of the two kinds.
2. Line 150-155, authors generated initial PCGs labels via RFM and GEE. Did the authors use some manually labeled data to validate the results?
3. Line 100-105, this paper mentioned a case which used commercial satellite data to extract the PCGs. What's its resolution? And what’s different from that research? The authors point that the scientific problem is how to use open access data to generate PCGs map. I think this was not the true scientific problem.
4. Figure 4 illustrated the multiple-temporal NDVI profile of bareland, PCGs and PMFs in a representative sub-region of Gansu Province, China. There are PMFs in some southern area. Farmer used plastic much earlier than the northern area, which would induce the difference of multiple-temporal NDVI.
5. Line 420-425, the PCGs dataset from the university of Copenhagen is the year of 2019. Differences between the two data years may result in discrepancies.
Of course, it would be better if the data include a few more years.

Citation: https://doi.org/10.5194/essd-2024-538-CC1
- AC4: 'Reply on CC1', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC4
CC2:
'Comment on essd-2024-538', Jie Bai, 05 Jun 2025

This manuscript developed a novel approach for producing the global 10-meter Plastic-covered greenhouse (PCGs) dataset for 2020 year. This approach combines the active learning strategy and the deep learning model, so as to let the model to learn more robust PCG features by optimizing weak labels. The results of the PCGs classification has been compared with publicly released land use and land cover (LULC) datasets, and showed the high accuracy. This PCGs dataset can characterize the global spatial distribution of plastic-mulched coverage in 2020. This manuscript is novel in topic selection, innovative in technical approach, and solves the problem of the lack of global-scale plastic-mulched products, which is of high scientific significance and practical value.
The following issues still need to be revised.
(1) Do the samples include actual ground-collected sample points? The manuscript has examples from Gansu Province in China, are there similar ground samples from other areas?
(2) From table 1. we can see that the UA of Non-PCG is as high as 99.99%, and the that of PCG is 84.18%. Do all these sample points for accuracy evaluation also come from GEE automatically selected or manually decoded? Is there a relationship between the high accuracy of Non-PCG and the sample selection?
(3) In the section of ‘4.3 Comparison with other studies’, the authors performed a comparison of the results of the spatial extraction of PCG in different regions of the globe. In addition, it is advisable to compare the total area of PCG by different continents (or regions). For example, the total areas of PCG in a region acquired by different data products, or compare the proportion of area where PCG overlaps in the same region by different products. This part of the study needs to be deepened in terms of difficulty, and merely comparing spatially with the LULC products or similar PCG products does not seem to be enough to prove the accuracy of this product. It is also recommended to add some statistical yearbooks or public data information for comparison.

Citation: https://doi.org/10.5194/essd-2024-538-CC2
- AC3: 'Reply on CC2', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC3
RC2:
'Comment on essd-2024-538', Anonymous Referee #2, 16 Jun 2025

The manuscript presents the development of a global plastic greenhouse distribution map (Global-PCG-10) based on Sentinel-2 10 m imagery. Weak labels were generated via a random forest classifier, and these were refined through a deep learning framework integrated with an active learning strategy. The spatial patterns, extent, and proportional coverage of global PCGs were analyzed, yielding results of clear scientific merit and practical utility. However, several issues and deficiencies must be addressed prior to publication:
1. Some in-text citations do not conform to the journal’s formatting requirements. For example, in line 64 the citation “Aguilar et al., (2016)” incorrectly includes a comma. Please review and standardize all reference notations.
2. In Figure 2 (Phase 2) the train/validation split is stated as 7:3, whereas in the main text it is described as 8:2. Please verify and ensure consistency in both places.
3. In lines 162-163 authors state that the initial labels were generated via GEE and RF classification. Was any quality control applied to these labels (e.g., manual verification rate, error‐rate assessment)? Please clarify.
4. In lines 165-166 (Stage 3) mention “post-processing was applied to the PCGs classification results to eliminate isolated noises” without specifying the algorithm or parameter settings. Please provide details of the post-processing method so that the results are reproducible.
5. The manuscript alternates between “PCG” and “PCGs” (e.g., line 166 “PCGs classification results” vs. Figure 2 “PCG classification results”). Please unify the terminology throughout.
6. In lines 239-243 outline the overall network architecture but omit implementation details of the MDCN and non-local modules. Schematic diagrams or detailed descriptions of these submodules are required.
7. Section 3.2.2 provides an overview of the active‐learning strategy but lacks specifics on the number of iterations, per-iteration sample‐selection criteria, and stopping conditions. Please supply quantitative stopping rules rather than the current qualitative phrase “until performance stabilizes or results are satisfactory.”
8. In lines 253-254, authors state that “if results are unsatisfactory, significantly altered initial labels are selected for further training,” but “expected standards” and “significant alteration” are not quantified. Please define the thresholds and the human‐involvement workflow in detail.
9. In lines 270–275, authors note the use of the Adam optimizer and data partitioning but omit critical hyperparameters such as total epochs, batch size, and learning‐rate decay strategy. Please provide these training details.
10. In lines 272-273, the dataset is described as comprising 14 825 training images and 3 707 validation images. Please explain how the sample was divided (randomly or by region? Is global diversity guaranteed?).
11. In line 292, the term “false positives (FP)” is incorrectly written as “false negatives.” This typo may confuse the metric definitions. Please verify that all terminology and formulas in the text and tables are consistent.
12. Section 4.2 presents only examples with no significant false alarms, which seems inconsistent with the overall recall rate of 84%. Please include error‐case analysis and discuss the causes of missed detections.
13. In lines 374-375, authors apply balanced sampling between PCG and non‐PCG in the test set, which does not reflect their real-world scarcity and may lead to overestimated performance. Please discuss the impact of balanced sampling on evaluation or provide results under the true class distribution.
14. In lines 381-382 the confusion matrix shows that the PCG class has a UA of 84.18% (FN = 3 243), yet no discussion of the false-negative causes is provided. Please analyze the reasons for these missed detections.
15. In lines 381-382, non‐PCG exhibits UA = 99.99% and PA = 86.30%, while PCG shows UA = 84.18% and PA = 99.99%, indicating a strong asymmetry in false‐positive and false‐negative distributions. Please discuss the origin of this bias.
16. In section 4.3.2, authors compare visually against Tong et al. (2024) using 3 m data (Figure 12) but lack objective quantitative metrics. Please add quantitative evaluation and report specific comparative values.
17. Recommend including a discussion of potential applications and the limitations of the produced dataset.

Citation: https://doi.org/10.5194/essd-2024-538-RC2
- AC2: 'Reply on RC2', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC2

Status: closed

RC1:
'Comment on essd-2024-538', Anonymous Referee #1, 16 Jan 2025

Global-PCG-10: a 10-m global map of plastic-covered greenhouses derived from Sentinel-2 in 2020
Manuscript number: essd-2024-538
General comments.
This very interesting paper use machine learning and deep learning on Sentinel-2 10 m GSD images to obtain a global map of plastic-covered greenhouses (PCGs). Really, it is not the first global PCG maps since Tong et al. (2024) already published other global PCG map derived from PlanetScope images, a commercial satellite with 3 m GSD, but using also in the first steps Sentinel-2. Although both works have a similar objective (i.e., to attain a global PCG map), the strategies used were quite different.
Thanks to the attained global PCG map, Niu et al. (2025) give interesting data about the area of PCG around the world, the major concentrations, spatial distribution, etc.
The manuscript is well written and it is worth being published. However, a few specifics comments should be taken into account.
Specific comments.
- Some cites in the manuscript appears with an extra comma. For example, in Page 2, Line 64, the cites “Aguilar et al., (2016) and Yang et al., (2017) independently developed…” should be “Aguilar et al. (2016) and Yang et al. (2017) independently developed…” Similarly, “Zhang et al., (2022a)” in Page 2, Line 66, should be “Zhang et al. (2022a)”. Please, correct this issue throughout the manuscript.
- Page 3, Line 71. The cite (Zhang et al., 2024) should be (Zhang et al., 2024a).
- Page 3, Line 76. Zhang et al., 2023 should be 2023a or 2023b. Please, review it.
- Page 5, Line 139-140. “Actually, Sentinel-2 is a constellation consisting of two satellites, i.e., Sentinel-2A and Sentinel-2B, which are in the same sun-synchronous orbit while phased at 180° to each other”. In fact, there is a new Sentinel-2C. You should speak a little about it.
- Page 6, Line 163. In Figure 2 (Stage 2) the train/validation ratio is 7:3, and in the manuscript you wrote 8:2. Is it a mistake in the Figure?
- Page 7, Line 175. In the caption of Figure 3, you should clarify that the size of the reference samples (512×512) are pixels and not meters.
- Page 9, Line 200. In the caption of Figure 4, it is written Multiple-temporal NDVI. Is not more appropriated multi-temporal NDVI?
- Page 9, Line 205. “1> Spectral features”. Strange login method.
- Page 10, Line 212. “2> Textural features”. Strange login method.
- Page 11, Line 237. You should clarify also in the manuscript that the size of the reference samples (512×512) are pixels and not meters.
- Page 12, Line 273. In Figure 2 (Stage 2) the train/validation ratio is 7:3, and in the manuscript you wrote 8:2. Please review it.
- Page 13, Line 286. Fu et al. (2021) is not in reference section.
- Page 15, Line 332-335. There are some numbers without thousands separation (e.g., 9874.51 km2, 2530.56 km2, 8224.90 km2).
- Page 16, Line 344. Figure 8a is not cited in the manuscript, and it should be.
- Page 18, Line 375. Why 20500 points for GH and 20500 for Non-GH. Justify this figure.
- Page 19, Line 381. Table 1 shows the confusion matrix where OA, User Accuracy (UA) and Producer Accuracy (PA) are depicted. Really, UA=Recall and PA=Precision, so, Table 2 is not necessary. The only data useful in Table 2 is F1 Score. I think that you should rewrite the methods and results about the accuracy assessment. Furthermore, Why is the classification so biased? For example, UA is 99.99% and PA is 86.30% for Non-GH and, UA is 84.18% and PA is 99.99% for GH.
- Page 22, Line 421-422. “… and in May 2024, the University of Copenhagen published a global 3-m PCGs dataset also in 2019”. Please, you should cite Tong et al. (2024) here.
- Page 22, Line 430. You should cite Tong et al. (2024) properly in the caption of Figure 12.
- Page 23, Line 434. “Tong et al., (2024) acquired from 3-m …”. Again, this cite appears with an extra comma.
- Page 30, Line 667. “Zhang, X., Liu, L., and Chen, X.: Global annual wetland dataset Data Descriptor at 30 m with a fine classification system from 2000 to 2022, Sci. Data, https://doi.org/10.1038/s41597-024-03143-0, 2024c”. This reference do not appear in the manuscript.
Final Comments:
It is very important that the global PCG map and the code are accessible to researchers. I have tested that the code for generating the initial labels of PCGs is publicly available via the following link on Google Earth Engine: https://github.com/MrSuperNiu/Greenhouse_Classification_GEE. It consists of feature extraction, RF classification, etc. Additionally, the code of APC-Net is accessible through the following link: https://github.com/MrSuperNiu/APCNet. The Global-PCG-10 dataset is stored on figshare, and can be downloaded here: https://doi.org/10.6084/m9.figshare.27731148.v2 (Niu et al., 2024).

Citation: https://doi.org/10.5194/essd-2024-538-RC1
- AC1: 'Reply on RC1', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC1
CC1:
'Comment on essd-2024-538', Liang He, 06 Feb 2025

This article used deep learning methods to produce a refined distribution of global Plastic-covered greenhouse in 2020. The methodology of the paper is sound and the data is novel. Although there are some global PCG data in the current study, the data set has some value in terms of resolution. There is some value for global agricultural macro-management.

There are some problems and deficiencies that need to be corrected as follows:
1. Plastic-covered greenhouse can divided into daylight greenhouses and plastic greenhouses. Is there a significant difference in the spectra of the two kinds.
2. Line 150-155, authors generated initial PCGs labels via RFM and GEE. Did the authors use some manually labeled data to validate the results?
3. Line 100-105, this paper mentioned a case which used commercial satellite data to extract the PCGs. What's its resolution? And what’s different from that research? The authors point that the scientific problem is how to use open access data to generate PCGs map. I think this was not the true scientific problem.
4. Figure 4 illustrated the multiple-temporal NDVI profile of bareland, PCGs and PMFs in a representative sub-region of Gansu Province, China. There are PMFs in some southern area. Farmer used plastic much earlier than the northern area, which would induce the difference of multiple-temporal NDVI.
5. Line 420-425, the PCGs dataset from the university of Copenhagen is the year of 2019. Differences between the two data years may result in discrepancies.
Of course, it would be better if the data include a few more years.

Citation: https://doi.org/10.5194/essd-2024-538-CC1
- AC4: 'Reply on CC1', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC4
CC2:
'Comment on essd-2024-538', Jie Bai, 05 Jun 2025

This manuscript developed a novel approach for producing the global 10-meter Plastic-covered greenhouse (PCGs) dataset for 2020 year. This approach combines the active learning strategy and the deep learning model, so as to let the model to learn more robust PCG features by optimizing weak labels. The results of the PCGs classification has been compared with publicly released land use and land cover (LULC) datasets, and showed the high accuracy. This PCGs dataset can characterize the global spatial distribution of plastic-mulched coverage in 2020. This manuscript is novel in topic selection, innovative in technical approach, and solves the problem of the lack of global-scale plastic-mulched products, which is of high scientific significance and practical value.
The following issues still need to be revised.
(1) Do the samples include actual ground-collected sample points? The manuscript has examples from Gansu Province in China, are there similar ground samples from other areas?
(2) From table 1. we can see that the UA of Non-PCG is as high as 99.99%, and the that of PCG is 84.18%. Do all these sample points for accuracy evaluation also come from GEE automatically selected or manually decoded? Is there a relationship between the high accuracy of Non-PCG and the sample selection?
(3) In the section of ‘4.3 Comparison with other studies’, the authors performed a comparison of the results of the spatial extraction of PCG in different regions of the globe. In addition, it is advisable to compare the total area of PCG by different continents (or regions). For example, the total areas of PCG in a region acquired by different data products, or compare the proportion of area where PCG overlaps in the same region by different products. This part of the study needs to be deepened in terms of difficulty, and merely comparing spatially with the LULC products or similar PCG products does not seem to be enough to prove the accuracy of this product. It is also recommended to add some statistical yearbooks or public data information for comparison.

Citation: https://doi.org/10.5194/essd-2024-538-CC2
- AC3: 'Reply on CC2', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC3
RC2:
'Comment on essd-2024-538', Anonymous Referee #2, 16 Jun 2025

The manuscript presents the development of a global plastic greenhouse distribution map (Global-PCG-10) based on Sentinel-2 10 m imagery. Weak labels were generated via a random forest classifier, and these were refined through a deep learning framework integrated with an active learning strategy. The spatial patterns, extent, and proportional coverage of global PCGs were analyzed, yielding results of clear scientific merit and practical utility. However, several issues and deficiencies must be addressed prior to publication:
1. Some in-text citations do not conform to the journal’s formatting requirements. For example, in line 64 the citation “Aguilar et al., (2016)” incorrectly includes a comma. Please review and standardize all reference notations.
2. In Figure 2 (Phase 2) the train/validation split is stated as 7:3, whereas in the main text it is described as 8:2. Please verify and ensure consistency in both places.
3. In lines 162-163 authors state that the initial labels were generated via GEE and RF classification. Was any quality control applied to these labels (e.g., manual verification rate, error‐rate assessment)? Please clarify.
4. In lines 165-166 (Stage 3) mention “post-processing was applied to the PCGs classification results to eliminate isolated noises” without specifying the algorithm or parameter settings. Please provide details of the post-processing method so that the results are reproducible.
5. The manuscript alternates between “PCG” and “PCGs” (e.g., line 166 “PCGs classification results” vs. Figure 2 “PCG classification results”). Please unify the terminology throughout.
6. In lines 239-243 outline the overall network architecture but omit implementation details of the MDCN and non-local modules. Schematic diagrams or detailed descriptions of these submodules are required.
7. Section 3.2.2 provides an overview of the active‐learning strategy but lacks specifics on the number of iterations, per-iteration sample‐selection criteria, and stopping conditions. Please supply quantitative stopping rules rather than the current qualitative phrase “until performance stabilizes or results are satisfactory.”
8. In lines 253-254, authors state that “if results are unsatisfactory, significantly altered initial labels are selected for further training,” but “expected standards” and “significant alteration” are not quantified. Please define the thresholds and the human‐involvement workflow in detail.
9. In lines 270–275, authors note the use of the Adam optimizer and data partitioning but omit critical hyperparameters such as total epochs, batch size, and learning‐rate decay strategy. Please provide these training details.
10. In lines 272-273, the dataset is described as comprising 14 825 training images and 3 707 validation images. Please explain how the sample was divided (randomly or by region? Is global diversity guaranteed?).
11. In line 292, the term “false positives (FP)” is incorrectly written as “false negatives.” This typo may confuse the metric definitions. Please verify that all terminology and formulas in the text and tables are consistent.
12. Section 4.2 presents only examples with no significant false alarms, which seems inconsistent with the overall recall rate of 84%. Please include error‐case analysis and discuss the causes of missed detections.
13. In lines 374-375, authors apply balanced sampling between PCG and non‐PCG in the test set, which does not reflect their real-world scarcity and may lead to overestimated performance. Please discuss the impact of balanced sampling on evaluation or provide results under the true class distribution.
14. In lines 381-382 the confusion matrix shows that the PCG class has a UA of 84.18% (FN = 3 243), yet no discussion of the false-negative causes is provided. Please analyze the reasons for these missed detections.
15. In lines 381-382, non‐PCG exhibits UA = 99.99% and PA = 86.30%, while PCG shows UA = 84.18% and PA = 99.99%, indicating a strong asymmetry in false‐positive and false‐negative distributions. Please discuss the origin of this bias.
16. In section 4.3.2, authors compare visually against Tong et al. (2024) using 3 m data (Figure 12) but lack objective quantitative metrics. Please add quantitative evaluation and report specific comparative values.
17. Recommend including a discussion of potential applications and the limitations of the produced dataset.

Citation: https://doi.org/10.5194/essd-2024-538-RC2
- AC2: 'Reply on RC2', Q. L. Feng, 30 Jul 2025
  
  Thank you very much for your helpful comments. The manuscript has been fully revised according to your suggestions. The point-to-point response to your comments is in the attachment, and responses are in blue.
  
  Citation: https://doi.org/10.5194/essd-2024-538-AC2

Supplement

https://doi.org/10.5194/essd-2024-538-supplement

Data sets

Global-PCG-10: a 10-m global map of plastic-covered greenhouses derived from Sentinel-2 in 2020 Bowen Niu, Quanlong Feng, Bingwen Qiu, Shuai Su, Xinmin Zhang, Rongji Cui, Xinhong Zhang, Fanli Sun, Wenhui Yan, Siyuan Zhao, Hanyu Shi, Cong Ou, Xiaolu Yan, Jianhua Gong, Gaofei Yin, Jianxi Huang, Jiantao Liu, Bingbo Gao, Xiaochuang Yao, Jianyu Yang, and Dehai Zhu https://doi.org/10.6084/m9.figshare.27731148.v2

Viewed

Total article views: 1,439 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
1,169	232	38	1,439	48	20	30

HTML: 1,169
PDF: 232
XML: 38
Total: 1,439
Supplement: 48
BibTeX: 20
EndNote: 30

Views and downloads (calculated since 09 Jan 2025)

Month	HTML	PDF	XML	Total
Jan 2025	235	74	6	315
Feb 2025	138	14	1	153
Mar 2025	103	14	1	118
Apr 2025	112	19	2	133
May 2025	168	19	4	191
Jun 2025	138	29	11	178
Jul 2025	125	34	12	171
Aug 2025	150	29	1	180

Cumulative views and downloads (calculated since 09 Jan 2025)

Month	HTML	PDF	XML	Total
Jan 2025	235	74	6	315
Feb 2025	138	14	1	153
Mar 2025	103	14	1	118
Apr 2025	112	19	2	133
May 2025	168	19	4	191
Jun 2025	138	29	11	178
Jul 2025	125	34	12	171
Aug 2025	150	29	1	180

Viewed (geographical distribution)

Total article views: 1,435 (including HTML, PDF, and XML) Thereof 1,435 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 28 Aug 2025

Download

Preprint (9022 KB)
Metadata XML

Short summary

We have proposed a novel framework to generate the firstly publicly released 10-m global PCGs map in 2020 derived from Sentinel-2. Results show that the global PCGs area is approximately 14,259.85 km² in 2020, mainly distributed between 30° N and 40° N. China, not only has the largest area of PCGs in Asia but also ranks first worldwide, with a PCGs area of 8,224.90 km², making up 57.67 % of the global and 83.29 % of the Asia. The overall accuracy of our Global-PCG-10 is satisfactory of 92.08 %.


Total:	0
HTML:	0
PDF:	0
XML:	0