GCI30: a global dataset of 30&thinsp;m cropping intensity using multisource remote sensing imagery

Zhang, Miao; Wu, Bingfang; Zeng, Hongwei; He, Guojin; Liu, Chong; Tao, Shiqi; Zhang, Qi; Nabil, Mohsen; Tian, Fuyou; Bofana, José; Beyene, Awetahegn Niguse; Elnashar, Abdelrazek; Yan, Nana; Wang, Zhengdong; Liu, Yiliang

doi:https://doi.org/10.5194/essd-13-4799-2021

Articles | Volume 13, issue 10

https://doi.org/10.5194/essd-13-4799-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/essd-13-4799-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 10

Data description paper

|

21 Oct 2021

Data description paper |

| 21 Oct 2021

GCI30: a global dataset of 30 m cropping intensity using multisource remote sensing imagery

Miao Zhang, Bingfang Wu, Hongwei Zeng, Guojin He, Chong Liu, Shiqi Tao, Qi Zhang, Mohsen Nabil, Fuyou Tian, José Bofana, Awetahegn Niguse Beyene, Abdelrazek Elnashar, Nana Yan, Zhengdong Wang, and Yiliang Liu

Download

Final revised paper (published on 21 Oct 2021)
Supplement to the final revised paper
Preprint (discussion started on 30 Jun 2021)
Supplement to the preprint

Interactive discussion

Status: closed

CC1:
'Comment on essd-2021-86', J. Fan, 11 Aug 2021

This is a great study in support of the global food secuity. The increasing polulation on earth gives rise to the increasing demand for food supply. It should be an effective way to increase the food production by the adequate use of land resources, for example, increasing the cropping intensity. The multiple cropping system is a major farming practice in some countries in the world, in particular in Asia. Therefore the monitoring of the cropping intensity at 30 meter resolution may provide cropping intensity information of close to the field level that is the unit of calculating the cropping index in agronomy community.
In this article, I'd like to raise a question to the authors. The question is that the cropping season is not completely in line with the calendar year. In some part of the world, the crop growing season of winter crops crosses the calendar year. How did you deal with this situation when you retrive the cropping/non croping information with the time series of satellite data.

Citation: https://doi.org/10.5194/essd-2021-86-CC1
- AC1: 'Reply on CC1', Miao Zhang, 23 Sep 2021
  
  The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2021-86/essd-2021-86-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/essd-2021-86-AC1
CC2:
'Comment on essd-2021-86', J. Fan, 11 Aug 2021

Dear Authors,
I'd like to raise another question after I downloaded a couple of tiles of data. The whole image was devided into 504 tiles that seems too many to search and download. Is it possible to aggregate some tiles into the continent level and then there are only less 10 large images. You may consider to keep 504 tiles and add new continent level images online. I assume that the latter approach will facilitate the user to download the images for large area study.
Thanks
Jinlong Fan

Citation: https://doi.org/10.5194/essd-2021-86-CC2
- AC2: 'Reply on CC2', Miao Zhang, 23 Sep 2021
  
  Thanks for Dr. Fan’s comment and suggestions on optimizing the files provided on the data sharing platform. We agree with the reviewer that 504 tiles are too much for the users to find specific tile or tiles over the area of interests.
  Actually, before open discussion of our manuscript, the Topical Editor also gave a comment on this. As suggested by the Topical Editor, we added three files in the repository. One is an index map in jpg format with labels to indicate the indices of the tiles available at https://dataverse.harvard.edu/file.xhtml?fileId=4771683&version=2.0. We also provide a shapefile so that the users could overlap the polygons of all tiles with their own study areas to identify which tiles of data are in need. In addition, we provided readme document in the repository.
  Based on the index map and polygons provided, users could not only access to the data, but also can easily find the data they want through the index map as suggested by the topical editors. We also agree with the suggestion of aggregating some tiles into the continent level providing limited number of files with large data volume for each. We aggregated the tiles accordingly and even packaged the data for each continent and uploaded the .zip files but failed because file upload limit is 2.5 GB per file by Harvard Dataverse where the data is shared. If the users want to download the images for large area study, we recommend using the bulk download option Harvard provided.
  
  Citation: https://doi.org/10.5194/essd-2021-86-AC2
RC1:
'Comment on essd-2021-86', Liang You, 12 Aug 2021

Please see the attached file

Citation: https://doi.org/10.5194/essd-2021-86-RC1
- AC4: 'Reply on RC1', Miao Zhang, 23 Sep 2021
  
  We would like to thank Dr. You for the constructive suggestions that help significantly improve the research and the quality of this work. We revised the manuscript according to these comments, and we provided our detailed responses to all the points raised in the supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-86-AC4
RC2:
'Comment on essd-2021-86', Anonymous Referee #2, 21 Aug 2021

This study combined Landsat, Sentinel-2, and MODIS images to generate a global cropping intensity map at a spatial resolution of 30 meters between 2016 and 2018. This study also validated the cropping intensity map using thousands of pixels from time series satellite images and currently available PhenoCam data. Then this study compared the resultant cropping intensity map with other cropping intensity datasets. This is nice work. The writing and the logic of this manuscript are good. However, I have some comments which may help the authors improve this study.

Page 1 Line 34-35: This study did not do the relevant analysis to support this conclusion.

Page 2 Line 14: Here is the definition of cropping intensity. What is the continuous cropping type you mentioned in the manuscript?

Page 3 Line 6: You may need to highlight why you combined Landsat, Sentinel-2 and MODIS data somewhere in the Introduction section.

Page 3 Line 15-25: According to your Google Earth Engine code, this study included forest, water, and urban mask to help integrate the cropland layer. But I did not find any description in this section.

Page 4 Line 4-5: Why not using surface reflectance data?

Page 4 Line 16-17: Landsat and Sentinel-2 are TOA data, and MODIS data are surface reflectance (SR) data if my understanding is right. The reflectance values from TOA data and SR data should be very different, especially when including the blue band to calculate EVI. Which version of MODIS data are used?

Page 5 Line 1-2: More details are needed for the data gap filling and smoothing. The integrated data is 30 meters, but what is the temporal resolution?

Page 5 Line 14: what is the size for these samples?

Page 5: Line 25: How did you separate flooded and non-flooded croplands?

Page 6 Line 1: Why did you choose the 50% of the NDVI amplitude?

Page 7 Figure 2: This figure is not clear to me. Could you please the ticks for the x-axis?

Page 8 Line 9-25: A brief introduction of these products is needed. Or you can move Table 2 here and add the two phenology datasets in the table.

Page 9 Line 14-25: Could you explain why the underestimation and overestimation happened based on those samples (RDsat)?

Page 10 Figure 4: I could not see the cropping cycles dots.

Page 11 Figure 5: The colors for different cropping intensity types are too close. Please modify the colors.

Page 14 Line 2: Why excluding the continuous cropping pixels?

Page 14 Line 19-20: Why “Wu et al(2018) might overestimate the annual harvest area and …”?

Page 14 Table 2: Please cite the relevant references. FAO data in 2010 is used for comparison, and the FAO data in 2016-2018 should be better for comparison.

Page 15 Line 7: MDC12Q2 should be MCD12Q2.

Page 16 Line 4: This study “provides insight only into the current actual cropping intensity (Page 17 Line 18)” instead of “agricultural land use management”.

Page 16 Line 7: I agree that the GCI30 generated by this study reduces uncertainties caused by the mixed pixel effect. However, suppose you would like to say you generated a more accurate global CI estimation. In that case, you need to do accuracy assessment for each cropping intensity product based on their specific definition and the same set of ground reference samples.

Table S1: You may add the spatial resolution for each product.

This manuscript is too long and could be shortened.

Citation: https://doi.org/10.5194/essd-2021-86-RC2
- AC5: 'Reply on RC2', Miao Zhang, 23 Sep 2021
  
  We would like to thank the Referee for the constructive suggestions that help significantly improve the research and the quality of this work. We revised the manuscript according to these comments and provided our detailed responses point to point to address all the issues raised. Please check the detailed responses in the supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-86-AC5
AC3:
'Comment on essd-2021-86 from Topical Editor', Miao Zhang, 23 Sep 2021

Dear Authors,
I still have two comments for #5 and #6. Regarding #5, it is not uncertainty. It sounds like quality indicator. Regarding #6, my suggestion is to add a table of CI statistics from different products. It will be of great help to provide such table at the country level in SI. I will send it out for review now to speed the process. But please address these two comments together with comments from reviewers when you have them and if you receive a decision of revision.
#5. The uncertainty or quality of the data should be added in the product. As seen from this paper, the spatial variation of uncertainty is significant.
#6. A result table of comparison with other studies can be added. It is difficult to evaluate the difference of this product with other studies based on current presentation.

Citation: https://doi.org/10.5194/essd-2021-86-AC3
- AC6: 'Reply on AC3', Miao Zhang, 24 Sep 2021
  
  We would like to thank for the constructive comments and suggestions from Topical Editor that help significantly improve the research and the quality of this work. We further revised the manuscript addressing the two comments, and we provided our detailed responses to your concerns in the supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-86-AC6

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Miao Zhang on behalf of the Authors (24 Sep 2021) Author's response Author's tracked changes Manuscript

ED: Publish subject to technical corrections (27 Sep 2021) by Yuyu Zhou

ED: Publish as is (29 Sep 2021) by Yuyu Zhou

AR by Miao Zhang on behalf of the Authors (30 Sep 2021) Author's response Manuscript

Download

The requested paper has a corresponding corrigendum published. Please read the corrigendum first before downloading the article.

Article (6873 KB)
Full-text XML

Short summary

Cropping intensity (CI) is essential for agricultural land use management, but fine-resolution global CI is not available. We used multiple satellite data on Google Earth Engine to develop a first 30 m resolution global CI (GCI30). GCI30 performed well, with an overall accuracy of 92 %. GCI30 not only exhibited high agreement with existing CI products but also provided many spatial details. GCI30 can facilitate research on sustained cropland intensification to improve food production.