General comments :

The preprint is dealing with plasmaspheric plumes in the Earth’s magnetosphere and their occurrence rate dependence to geomagnetic storms intensity. The study is based on observations and simulations. The study is based on Van Allen Probe A observations of plasmaspheric plumes. Results are compared in detailed to the output of a previous study (Lee et al., 2016) that focused on the observations of plumes close to the magnetopause. Simulations of plume evolution are presented for two geomagnetic storms of different intensities to explain the different results between the present study and Lee et al. study. The overall impression of the study is positive: the paper is well-organized, the reasoning is logical, figures support the interpretation, and considering simulations to explain observations make the conclusion stronger.

Specific comments:

1. Only VAP-A observations are used (and not VAP-B), which make sense for the occurrence rate calculations. It is therefore surprising that only VAP-B orbit is overplotted on the simulation output with the indication “the black curves indicate the observed plasmaspheric plume” (Figure 5 and Figure 7). Why is that plume observation made by VAP-B? one could expect observations from VAP-A or, even better from both probes.

2. The main finding of the study is that the occurrence rate of plumes at lower l-shell is higher during large geomagnetic storms and higher at large l-shell during low geomagnetic storms. As shown in your simulations a given plume plume is larger close to the plasmapause (at low-shell)  that farther from the plasmapause (at larger L-shell). The plume criteria detection is however the same for every L-shells. How can this affect the occurrence rate estimation, and hence the conclusion?

3. The two simulated events correspond to geomagnetic storm of still relatively moderate intensity. Is there any reason for not choosing a Dst<-100nT storm for the second event?

L45: for EMIC the situation is a bit more complicated: EMIC are not preferentially observed in the high-density plumes (Usanova) excepted maybe for triggered emission (Grison)

L146: Is there any discontinuity in the density or the Sheeley model states a density of 5 at L=7?

Figure 4.c: It would be good to extend the time range to the stop time of the simulation (figure 5)

Technical corrections:

L25: A torus

L30: plasmaspheric particles: the convection given by the formula is sign-charge dependent

L72: deleted artificially: “discarded” might sound better

L94: define the “non storm” period

L101: dusk side: looking at the figure, you could even give a precise range 15-21MLT 

Plasmaspheric plumes play important roles in the inner magnetosphere. The current study statisticaly investigats the plumes observed by Van Allen Probes. Further more, it explains the difference of the observed features of plumes by Van Allen Probes and Cluster by performing test particle simulations. The simulations results explain well these different features. This is an interesting study and contributs to our understanding about plasmaspheric plumes. I suggest this preprint be published after minor revisions and list my comments and suggestions as follows:

A general question:during these intervals under study, where were Cluster satellites? Since authors compare observation results from Cluster and Van Allen Probes, is it possible to compare their observations during the same time intervals?

Line 55: ‘In this paper’, in situ measurements from Van Allen Probes are used to…

Line 73: what is the criteria for this ‘sharply’?

Line 102: But the time interval of them are shorter. Could you please normalize them and compare the occurance rate rather than simple number of events?

Line 114: It would be interesting to add MLT-L dependence. Maybe authors can plot ‘dial

‘ figures like in Figure 7, but color-code occurance rate in different MLT-L bins.

Line 115: How about storms with Dst lower than -70 nT?

Line 125: delete ‘disparity’ after ‘rates’

Line 127: test particle simulation’s’

Line 146: is the 5cc set up for the initial condition?

Line 166: I suggest add labels indicating some L values in the figure

Line 168: remove ‘in’ after ‘on’

Line 169: be lose → lost

Line 177: 40th hr (not shown)

Lines 179-182: This sentence is too long. I suggest authors to finish a sentence after ‘complicated’ on line 181 and to start a new sentence afterwards.

Line 203: ‘The initial distribution of electron density’ is set up in the same way as … on 30 April, and is shown …

Line 212: loss ‘to’ the magnetopause

Line 213: Is the upflow of electrons also stronger in strong storms? This can be an uncertainty in your simulation studies since you don’t have the upflow process included in your simulations but they can be different for storms of different levels.

Line 226: Again, I suggest calculating occurrance rates, instead of simply comparing number of events.

Line 232: remove ‘appears’ after ‘plume’

Line 257: lost ‘to’ the magnetopause 

Lines 279: ‘these two’ factors make ‘the Van Allen Probes’

Figure 1: Considering this is a study related with storm/non-storm periods, I suggest authors to add panels in this figure to show related geomagnetic indices (e.g., Dst, Kp, AU, AL, AE), and to add verticl lines indicating storm phases and the start of the storm if this is a storm period.

Figure 7 caption: line 449: on → above