Firstly, I want to thank the authors for completing the work for the community. It can be challenging and time-consuming to work with multiple historical satellite missions. As outlined in the manuscript, each dataset typically has distinct characteristics, and special considerations or care are often needed to ensure high-quality analysis results. I acknowledge such a huge effort associated with this work.

In this study, the authors examine five heritage satellite radar datasets and Sentinel-1 data to identify surge onset and termination in Svalbard using methods from Leclercq et al. (2021) and Kääb et al. (2023). The results agree well with existing records and have updated our knowledge of surging glaciers in Svalbard by adding a few more events and clarifying previous identifications. The study also shows that both C-band and L-band can be used to identify surge events based on radar backscatter changes, which could be useful information for future radar missions. The manuscript is well prepared, with detailed descriptions, and the discussion offers interesting validations and perspectives on surge frequency and external forcings. I enjoyed reading it.

The manuscript is ready to be accepted by TC, in my opinion, but if the authors have time, the following comments may be considered before publication.

• What is the availability of the data generated by this work, especially the mosaicked annual radar maps and the maps of NDI for each glacier? It would be good to guide the readers with a data availability section.
• L302-304: The ENVISAT—RADARSAT-2 and ERS-2—RADARSAT-2 analyses seem to be the only cross-platform comparisons throughout the study, but I can’t find any results or discussion later in the manuscript. What is their performance in terms of surge detection?
• L469: For one simulation, are all surges drawn from the same normal distribution? Or is each surge drawn from a different normal distribution with the mean and standard deviation randomly assigned? How many surge events are drawn?
• L491-498: It would be helpful to provide more details for readers to understand the simulation. According to Figure 16, this analysis is only performed for the case with F = 15/decade, correct? How many surging glaciers are there in one run? Are there also a few million simulations aggregated into these results?
• Copyediting suggestions:
  • L244: … detection over Svalbard “using” the ENVISAT…?
  • Figure 11: Since the glaciers are not aligned along the southwest to northeast direction, it would be good to add labels of glacier names to each surging glacier for better identification with the caption.