We thank the reviewer for their positive comments on our manuscript.

We have taken to heart the reviewer's comment that each of the independent observations of lightning occurrence that we compare with WGLC are based on different technology and ultimately measure different quantities. In our revision of the manuscript, we will stress this point. In particular, the comparison between WGLC and LIS/OTD requires some more discussion.

We take note that previous studies showed that lightning flashes detected by LIS/OTD could be composed of multiple lightning strokes. Rudlosky and Shea (2013) found that, on average, WWLLN captured 1.5 strokes for each LIS/OTD flash, although 71.5% of the WWLLN-matched LIS/OTD flashes were from a single lightning stroke.

In our analysis, WGLC nearly always has lower lightning density than LIS/OTD. As noted above, a single LIS/OTD flash may be comprised of multiple lightning strokes. This implies that the detection efficiency of WWLLN may be even lower than might be assumed based on a simple comparison of the two datasets. Our analysis shows that WGLC is about three times more likely to detect LIS flashes over ocean (33%) than over land (10%).  Accounting for the mean multiple-stroke-per-flash discrepancy reported by Rudlosky and Shea (2013) implies that WGLC's implied detection efficiency using LIS/OTD as a standard would be 22% over ocean and 6.6% on land, which is in line with the 17.3% over ocean and 6.4% over land reported by Rudlosky and Shea (2013).

In our revised manuscript, we will include some further discussion on the issues comparing WGLC with LIS/OTD and the implications of the multiple-stroke-per-flash discrepancy on assessment of the absolute detection efficiency of WGLC. We also include the additional references helpfully provided by the reviewer.

As noted in this reviewer comment, not all of the data described in the manuscript was accessible at the PANGAEA link we provided in the original manuscript. We subsequently learned that PANGAEA does not accept updates to existing datasets, so we created a new submission that was initially processed on 23 March 2021. Unfortunately, PANGAEA has informed us that publishing of datasets takes up to 12 weeks, and so our updated submission has not yet been released with a new DOI.

This long delay in publishing our data on PANGAEA, and the reviewer's request for a daily version of the gridded data, prompted us to search for alternative solutions to making the WGLC dataset publicly available. To solve this problem and make our data immediately available, we have released the full dataset, including a daily-resolution version of the 30-minute global fields, on GitHub at this link:

https://github.com/ARVE-Research/WGLC

Going forward, we will either see if PANGAEA will accept another revision to our submission to accommodate the additional daily fields, or find another data publisher, e.g., zenodo, for the final version of the 2021 release of the WGLC. 

We thank the reviewer for their careful examination of our manuscript and dataset and will reply to the rest of the comments shortly.

We thank the reviewer for their overall positive view of our dataset, extensive comments, and eye to detail. These comments are constructive and will improve our revised manuscript. We respond to each of the reviewer comments below - reviewer's comments in italic and our responses in plain text.

1. The manuscript mentions that 5 arc-minute stroke density data (line 80) and stroke power data (line 129) are provided, but they are not available at the provided link ( https://doi.org/10.1594/PANGAEA.904253 ). Obviously this needs to be rectified/clarified before publication.

This comment has already been addressed in our previous response published above. The most up-to-date version of the WGLC dataset is available on Github, and if the manuscript is ready to be accepted before Pangaea publishes the dataset, we will find a different venue for publishing the dataset where it can immediately receive a DOI, e.g., figshare or zenodo.

2. Some of the language in the Abstract and Introduction is slightly over-the-top and or not supported by references. The dataset is valuable, I don’t believe that authors need to work so hard to persuade the reader of that. Specifics:

Ok, we are happy to tone down the language.

Line 1: “Lightning is one of the most important atmospheric phenomena” is rather subjective statement. I can imagine that many different atmospheric scientists might have different views on that, and that lightning wouldn’t feature in all of their “most important atmospheric phenomena” lists.

We'll go with "lightning is an important atmospheric phenomenon".

Line 18: “is the principle non-anthropogenic cause of wildfire ignitions”. Can the authors provide a reference where that has been quantified to back that up?

Interesting question. The relative rarity of other non-anthropogenic sources of wildfire ignitions make this kind of obvious: rolling rocks, volcanoes, meteor impacts, and spontaneous combustion occur infrequently, are local, and are seldom invoked as the cause of major wildfires, except possibly over geologic timescales, e.g., during the formation of Large Igneous Provinces. We consider windstorms and earthquakes that damage human infrastructure that lead to wildfire ignitions as "anthropogenic source" . We will clarify these points in our revised text and yes, we will try to find a reference. If the reviewer has a suggestion, we would be pleased to use it.

Line 22: “Large scale maps of lightning occurrence are as important as those for temperature or precipitation for many land surface (Hantson et al., 2016)”. Reading Hantson et al. 2016 reveals that not all fire-enabled vegetation models use lightning as an input, and beyond that not all land surface models even include a representation of fire (even if they arguably should). They all include temperature and precipitation (to my knowledge) so this statement in demonstrably false and should be rephrased.

Ok, we will explain that it is an important input for "some" vegetation-fire models.

Line 23: “… and atmospheric chemistry models (Finney et al., 2016)”. Following up the Finney et al 2016 paper, it does not appear that they state that maps of lightning importance are “as important as those for for temperature or precipitation”, and in fact they do not use lightning occurrence data at all. Rather, they study lightning parameterizations as used in atmospheric chemistry models and find a high correlation with temperature, which slightly works against the authors’ point. Again, please rephrase.

Likewise, we will qualify this statement adding that it is important for some models.

Line 25: “Observing and mapping lightning distribution at large spatial scales has thus been a priority for the community for nearly a century.” - a nice little factoid but not backed up with any reference. Maybe the Krider 2006 reference in the next paragraph should also apply, but on a cursory reading it doesn’t appear to refer to “large spatial scales”.

The first estimate of the global lightning flash rate was published 96 years ago (Brooks, 1925). Our apologies for omitting this citation, which will be added to the revision.

Brooks, C. E. P. (1925). The distribution of thunderstorms over the globe. Geophysical Memoirs, 3(24), 147-164. 

3. The spatial accuracy of WLLNN is quoted as 10 km (line 53) but also 3 km (line 351) and 2-3 km (line 393), please clarify.

At all places in the manuscript, we have clarified the horizontal accuracy of WWLLN to the network's published 3.4km (Rodger et al., 2005)

4. Whilst the main workflow (both WWLLN and WGLC) and details of the sensor network are well described, the methods by which the “deMaps” (line 105) are not described at all. It would helpful if the authors could describe this, however briefly. A link to a peer-reviewed publication would be adequate (there in no such publication behind the provided link, just data and code).

We performed minimal postprocessing of the DE maps - this consisted of linear interpolation to the target grid resolution - as part of this study. In our revised manuscript at the point where we describe the DE maps, we will add the citation to Hutchins et al. (2012) which details the theory behind and production of the DE maps.

5. On line 166 it is stated that “Strokes detected in the same gridcell and in the same hour were considered to be overlapping.” but within one gridcell and one hour there must have been at least two strokes (by virtue of the processing previously described). So how were the multiple strokes within that spatio-temporal frame matched to each other for purposes of determining exactly which strokes were missed and therefore what was the power of overlapping and non-overlapping strokes?

For our analysis of stroke power, we did not perform any filtering or set a detection threshold. We will clarify this point in the revised manuscript.

6. Figure 2 - the “DE” lines stop at 2018, but I don’t believe this is explained anywhere in the text. Please provide a sentence explaining this.

We will add an explanation to the figure 2 caption that we do not show the timeseries of uncorrected "A" WWLLN data after 2018 because we consider the postprocessed "AE" data to be more reliable and therefore use only the "AE" data in our gridded products. The "A" data shown in figure 2 are illustrative, to show the overall difference between these two sources.

7. Figure 9 - maybe a true log scale (or a smoother “faux log scale” such as Figure 7) would be more appropriate here. The sudden jump from 10 to 15 seems to introduce some visual artifacts and the color space is not currently well utilized with no purple and very little red on the plot. Same goes for Figure S6.

We will adjust the color scales for Figures 9 and S6 to better equalize the spatial coverage of each color on the map.

8. The statement on line 306 “here does not appear to be an overall spatial bias to the difference between datasets, i.e., the area of greatest anomaly shifts from year to year.” is very difficult to really assess given differences in overall amount of strokes and the quasi-log scale. I would suggest adding a plot of the year-by-year differences between the datasets after they have been normalised to each other (such that they have the same number of total strokes) to the supplementary if they really want to make this point. Then it should be immediately apparent if and how the area of greatest spatial bias moves around. It is also difficult to see how Figure S8 supports the statement “WGLC shows greater lightning density than NLDN along theGulf coast and in Southern Texas; this difference is most apparent in 2012” (line 309). It seems to be true for 2012 but it is hard to say for other years. The plot described above would clear this up.

In our revision we will include an additional supplementary plot showing the normalized differences between WGLC and NLDN. Indeed, these new figures will help the reader to see that the spatial distribution of differences between the two datasets is similar from year to year.

9. All map figures - longitude and latitude labels and tick marks are missing. Conventionally they should be included and it would aid detailed study of the plots (especially when comparing the maps to the zonal plots such as Figs 8 and S10 where latitudinal tick mark have been included).

We omitted tickmarks on most of the maps in the manuscript because we are aware that our maps will be reproduced in small size in the final version of the manuscript and we wanted to maximize the space available to show the map information. Nevertheless, we also see the reviewer's point particularly when comparing zonal plots with maps. We will add graticule tickmarks to the larger maps in our revision.

10. Is it not a bit misleading to include 2010-2012 in the climatological mean given the statement “Even with the adjustment for detection efficiency, the years 2010-2012 in the WGLC should be treated with caution,” (line 330)? Can the authors justify their inclusion of 2010-2012 in the climatology given that proviso?

Because we provide both the timeseries and the climatology, we allow the user to decide if they would prefer to make a climatology over a shorter period containing, perhaps, more reliable data. We choose to use as much of the data as possible in generating our climatology as there are places where lightning is rare, e.g., in the Arctic or over the oceans, that benefit from the extra years of observation when building the climatology, because even some positive lightning density, however small, is more realistic than zero values.

11. Line 343 - “Similar to their analysis that used LIS/OTD (Murray et al., 2012), we do not see any evidence of a global-scale relationship between WGLC lightning density and the multivariate ENSO index, total solar irradiance, or the QBO.” I don’t believe this is demonstrated anywhere in the manuscript. I am all in favor of reporting negative results, but there should be some evidence (correlations in the supplement or whatever) to flesh out this statement.

We will add supplementary figures to our revised manuscript where we compare WGLC lightning with ENSO, TSI, and the QBO.

12. Line 352 - “... with monthly resolution the current standard for most global gridded climate datasets (e.g., Fick and Hijmans, 2017; New et al., 2000; Wilson and Jetz, 2016).” - this statement does not really cut muster. Daily climate or sub-daily climate reanalysis dataset have been around for a “long time” in Earth system science and been combined with observational datasets to account for model biases, eg back in 2006 Princeton Hydrological Forcing Dataset.

Sheffield, J., G. Goteti, and E. F. Wood, 2006: Development of a 50-yr high-resolution global dataset of meteorological forcings for land surface modeling, J. Climate, 19 (13), 3088-3111

Consider also various incarnations of the CRUNCEP dataset as used here, for example:

Sitch, S. et al. Recent trends and drivers of regional sources and sinks of carbon dioxide. Biogeosciences 12, 653–679 (2015).

There are further examples of widely used climate datasets at daily (or finer) temporal resolution and half degree (or finer) spatial resolution (consider ERA5 for example).   The authors should either rephrase their reasoning or bite the bullet and provide the data at daily resolution (at least for the half degree dataset, in that case storage requirements would not be so enormous given that the current dataset is only 35.3 Mbytes compressed).

We see the reviewer's point and have taken this final suggestion to heart. As noted in our previous response to reviewer comment, we now release a daily version of the lightning timeseries as half-degree resolution, along with those data described in the original submission.