I thank the authors for their novel empirical investigation of thunderstorm conditions, and for their work to utilize the multi-parameter empirical information to reason about the meteorological dynamics. I think the manuscript is suitable for publication with minor revisions, and will be a helpful practical forecasting aid while supporting clear, physically-based forecast reasoning.

I wanted to comment on the choice of words for the two categories of lightning identified in the study.

“Wind” and “mass/moisture” could be made even more specific and descriptive. The word choice here matters because it conditions how the reader perceives the connection to what actually drives the cloud physics processes that can result in electrification. To this point, the authors note that “wind field” is, specifically, the synoptic-scale thermal wind (line 200), and implies quasi-geostrophic dynamics driving clouds formation. Likewise, mass/moisture is really the specific thermodynamics associated with conditional instability and upright moist convection. These more specific ideas are strongly implied by the authors' reasoning, and so for this reason I encourage the authors to consider adopting more specific terms.

Other comments specific to certain lines of the manuscript are below.

1. If the authors reduce the number of clusters to k=3, do the dark and light, red and blue markers combine more with each other than they do with the yellow markers?

2 Line 36: it is predominantly differential sedimentation rates, not atmospheric motion, that separates the charges after collision. After differential sedimentation, wind shear can act. Also, on line 212 and line 290-1 and line 345: a constant vertical velocity cannot separate charged hydrometeors; the separation has to come from sedimentation within the inertial frame, or by differential motion of particles from one inertial frame to another. A related point is that the cloud life time scale is much larger in winter storms, so even relatively slow vertical motions and charging rates could still result in sufficient charge for lightning.

3. Line 75: in k-means clustering, k is chosen a priori, but the text says it “yielded k=5 clusters,” which implies the the algorithm itself determined k and might be confusing to readers. The manuscript later clarifies how this k was chosen (sec. 3.3). I suggest changing “yielding” to “using.”

4. Line 110: "on the scenarios without lightning”: was the normalization done separately for the summer and winter populations, or aggregated across all no-lightning subsamples? It would also help to describe this process in order as a mathematical formula: first introduce the calculation of the mean and standard deviation of the no-lightning cases, and then state how the other variables were normalized using these quantities. Line 124 and 174 repeats this phrase, and with careful explanation here it can be removed there, unless I misunderstand and normalization is being applied several times.

5. Section 3.3: Were the PCA and the k-means methods applied independently? It seems so. Does independent application of the two methods help give confidence in the interpretation of the results? What were the key principles used to judge high- and low-meaning clusterings and PCA projections?

6. Line 243 and following: the definition of the cloud used to diagnose the relative cloud temperatures in summer and winter uses only droplets. Concerning the conclusion, “during lightning in winter clouds are – integrated over their depth – overall warmer than summer clouds,“ how would it change if both ice crystals and droplets were used together to define the cloud? Is there something special about droplets that makes them alone worth considering?

7. Line 310: HSLC storms (because some CAPE is present) typically do release upright convective instability even though they are during the cool season, so I appreciate the authors having introduced the HSNC idea here. It is not the month of the year but conditions associated with the synoptic pattern that control the mechanism by which lightning is generated.

Review of wcd-2021-68

Authors: Deborah Morgenstern et al.

Title: “Differentiating lightning in winter and summer with characteristics of wind-field and mass-field”

Recommendation: Acceptable pending major revisions

Overview

This manuscript explores the meteorological processes leading to lightning in winter and in summer in 

Northern Germany in an attempt to isolate conditions that differentiate between lightning and nolightning in summer and in winter. The authors use a pure data driven approach selecting cluster 

analysis and principal component analysis to group parameter derived from ERA5 reanalysis data into 

physically meaningful groups representing wind-field-dominated and mass-field dominated lightning 

conditions. 

While this manuscript is generally well written and informative, I have a number of suggestions that 

follow. Additionally, I have two major concerns that are highlighted below, which will require more 

thought and effort.

Major Comments

1) Northern Germany was chosen as study area to represent an area in the mid-latitudes and in 

flatland to minimize topographical triggering influences on lightning. The selected area is rather 

small. The results obtained in such a restricted area cannot be transferred to southern Germany 

nor to the rest of central Europe, which includes the Alpine region. Minimizing topographical 

triggering influences on lightning is a considerable restriction. Orography often acts as a trigger 

mechanism in summer as well as in winter. The limitations of the choice of the study area are 

not stated clearly in the manuscript. What would the results look like for other regions, e.g. 

southern Germany, the Alpine region or along the Mediterranean coast? Are the findings valid 

in other regions of (central) Europe? If not, how do they variate?

2) The study is limited to the two seasons winter (December, January and February) and summer 

(June, July and August). What about spring and autumn? How would the cluster analysis and 

principal component analysis deviate in these seasons from the presented seasons? Is it really 

necessary to group the data into seasons? If lightning occurs or not should not be defined by

the calendar but rather by the synoptic and atmospheric conditions that lead to the 

mechanisms of thunderstorm generation and the formation of lighting.

Minor/Grammatical Comments

Line 22: “copious amounts of moisture” – What type of moisture is meant here? Moisture near the 

ground or in the atmosphere? Moisture in form of specific humidity or relative humidity?

Line 54: Figure 1

Line 68: Although a .csv file is provided with a detailed description of the variables used in this study, I 

suggest to add a table with name, unit and meteorological category of the selected variables. 

Line 71: How are equally-sized subsamples formed? Randomly?

Lines 81-92: Can you provide a time series of all lightning that occurs within the selected domain? I 

assume there are years with higher lightning activities in the respective seasons? How were the 

“cell-hours” selected? Are they representative for an “average” season?

Lines 110-112: It is not entirely clear to me exactly how this is done. Please add more information or a 

concrete example.

Line 119: What does “yielding k=5 clusters” mean? 

Line 129: Use PCA as this was already introduced in line 120.

Figure 2: Variables derived from ERA5 reanalysis data are not observations. The labeled arrows are 

very hard to read. Could the readability of the figure be improved?

Figure 3: How would the stacked bar plot of the clusters differ for spring and autumn?

Line 148: Figure 2. Please be consistent with the usage of Figure x or (Fig. x) throughout the 

manuscript. E.g. use Figure in the text and Fig. in brackets.

Line 188: Within the mid-latitudes, there are many regions where lightning occurrence does not peak 

in summer months. Please check Taszarek et al. 2019. 

Line 274: Is the approach really purely data driven and completely independent of the season?

Line 285: What are “large amounts of CAPE” values? 

Line 295: Please check the order of the references throughout the manuscript. The years of the 

publications should be ascending.

Line 346: What are “substantial amounts of CAPE”?

References:

Taszarek, M., Allen, J., Púčik, T., Groenemeijer, P., Czernecki, B., Kolendowicz, L., Lagouvardos, K., 

Kotroni, V., & Schulz, W. (2019). A Climatology of Thunderstorms across Europe from a Synthesis of 

Multiple Data Sources, Journal of Climate, 32(6), 1813-1837. Retrieved Dec 5, 2021, 

from https://journals.ametsoc.org/view/journals/clim/32/6/jcli-d-18-0372.1.xml