See supplement with comments.

This study demonstrated a global irrigation dataset with 100 meters using irrigation performance during drought stress, which is a brand new way to detect irrigated and non-irrigated cropland. Furthermore, this MS finishes mapping the global central pivot irrigation system using the Deep Learning method. Also, it is interesting to detect special irrigation methods using deep learning methods. Overall, the MS was well-written and designed for readers. But there were still some concerns before this MS was accepted:

Major concerns:

1. About the resolution: In section 2.1 some coarse data was described as input data, but the final resolution of the irrigation map is 100 meters, so this will mislead some readers on how to produce a 100-meter irrigation map using 500-meter data.
2. About the IMZs: You mention that “65 MRUS in Copwatch served as the basis for further division of global cropland into 110 irrigation mapping zones (IMZs)”, what is the principle for further dividing these zones? Are these zones available or not?
3. About accuracy assessment: you collect many field points using the GVG app. How to distinguish irrigation field points during the field survey?
4. The irrigation map and GCPIS were identified using two ways (irrigation performance and DL), but some figures make me confused to display these two results.

Minor revision:

1. The preprocess of NDVI data in Line 160 should be further explained.
2. You could list some detailed maps of global CPIS in Figure 6 to make the global CPIS map clearer.
3. In Figure 16 it will be significant if the satellite images were added to give the reader a basis for their judgment.
4. IMZ was not so readable in Figure 1.
5. The English should be further polished and improved.

This manuscript introduced the GMIE-100 dataset, which identifies global irrigated cropland using drought stress performance and machine learning. This is a valuable dataset that could benefit various fields, including agriculture, environmental science, and water resource management. However, I have some major concerns about this MS that need the authors to clarify before it is further processed.

1 The title of the manuscript indicates that the dataset represents the largest irrigated area. How does the author interpret this "largest area"? This requires the author to provide explicit clarification within the text. Additionally, how does the author consider the possibility of overestimation of this largest area relative to the actual distribution, given that our focus is on the actual distribution range?

2 The samples are derived from different collection methods. It is crucial for the author to clarify whether samples collected through different methods exhibit consistent representation and describe irrigated land in the same manner. If their collection standards vary, the author needs to explicitly discuss the impact on the results.

3 In terms of accuracy assessment, merely providing overall accuracy is insufficient. Please refer to best practices for reporting accuracy as outlined in papers such as Olofsson et al. 2014 [1]. Moreover, I have not observed quantification of uncertainty, which necessitates further work from the author.

Olofsson P, Foody GM, Herold M, Stehman SV, Woodcock CE, Wulder MA. Good practices for estimating area and assessing accuracy of land change. Remote Sensing of Environment 2014; 148:42–57. https://doi.org/10.1016/j.rse.2014.02.015.

4 The results and discussion sections lack necessary citations. Many explanations proposed by the author lack corresponding literature support, which makes it difficult for me to be convinced of the correctness of your interpretations. Please see the annotations I've made in the manuscript.

5 I have made several annotations in the manuscript indicating areas that need revision. It is advised that the author make corresponding modifications and carefully review the entire document to rectify similar errors.