Interactive comment on “ Atmospheric rivers moisture transport from a Lagrangian perspective ” by A

This paper analyses atmospheric rivers (AR) across the North Atlantic and Europe (from NCEP 2 reanalysis from 1979 to 2012, 6-hourly time scale) using a Lagrangian perspective using the FLEXPART tool. I understand the interest of AR to explain some extreme events as flood or heavy snow, but I do not see what is the real added value of the approach and analyses in this paper. The introduction is not well organized (see minor points below) and the novelty of the analyses does not appear clear to me. My main concern is also that some of the conclusions are indeed very well known, for example that moisture in western Europe comes primarily from subtropical North C1114


Overall evaluation
The manuscript presents an analysis of the evaporation-minus-precipitation (E-P) signature along atmospheric river (AR) trajectories of hydrological relevance to various sectors of west Europe. The detection of AR is based on an existing method introduced in Lavers et al. (2012) with refinements to facilitate applications to different geographical sectors. Analysis of AR trajectories is based on a Lagrangian dataset produced by a global simulation of the FLEXPART model widely used in the community. The ERA-Interim reanalysis is used for the detection of ARs, and for forcing the FLEXPART model. The methods used are reasonable, and the results represent a useful contribution to the ongoing understanding (sometimes debate) of the moisture sources and transport associated with ARs.

C1122
I would like to recommend publication of the article in Earth Syst. Dynam. subject to major revisions suggested below.

AR detection
The detection of ARs is based on refinements to the method introduced in Lavers et al. (2012). Specifically, multiple (i.e., three) reference meridians are used instead of a fixed one for the entire west Europe. Landfall time and locations based on the three reference meridians are then pooled and regrouped into five sub-domains based on geographical relevance.
Part of the overarching difficulties involved in AR detection over large domains is the challenge to establish a universal threshold for the AR intensity, and the above is a potentially useful effort toward improved AR detection over large domains based on the Lavers et al. method, and may represent one of the novel aspects of the study.
However, it is not clear whether the refinements actually improved AR detection. For example, do ARs in the final five sub-domains better correlate with heavy precipitation in each sub-domain than does the original set of ARs based on a single reference meridian at 10W? I wish the authors would take the opportunity to show that the refined method indeed works better.

Trajectory analysis
To my knowledge the current study is one of the two studies that analyzed the E-P (or dq/dt) signature along AR trajectories, the other study being the case study in Stohl et al. (2008). In this regard, the current study is the first to present the E-P signature from a climatological perspective, a novel aspect not articulated in the paper currently.
What would make the paper more interesting and insightful would be to additionally analyze the E and/or P components of E-P to show the relative importance of E vs. P over different moisture source regions. The analysis, if done, would have important C1123 implications to observing and simulating ARs as precipitation is among the least well represented processes in GCMs which may limit our capability to realistically simulate the AR moisture balance along its trajectory.
What would usefully complement the E-P analysis would be the distribution of track densities, i.e., the count of parcels that contributed to the E-P calculations at each location, for example, see Figs. 3d-f of Rutz et al. (2015). With this information on track density the inference of AR moisture source regions would be more complete and compelling.
While I do not expect the authors to conduct all of the new analysis suggested in my major comments, I encourage the authors to take the opportunity to make the paper a more insightful and potentially more influential contribution to the science of ARs.

Minor comments and corrections
P2618L12: it would make more sense to give the parcel size; the total number of parcels is less relevant.