Summary:

This paper uses formulas predicting the isotopic composition of falling raindrops (delta_rain) as a function of drop size from Graf et al. 2017 and applies the model to raindrop size distributions measured during the ATOMIC campaign and compares the model predictions to ship-based rain collections in the region. The model is also informed by aircraft observations of atmospheric vapor stable water isotopes (delta_vapor) at cloud base and near-surface delta_vapor and relative humidity. The model-predicted delta_rain does not closely match the samples collected at the surface and the authors speculate why.

General comments:

1. The model formulation by Graf et al. 2017 has not been tested using aircraft observations of microphysics and atmospheric thermodynamic profiles to my knowledge. Previously observations of vapor and precip in these types of analysis were limited to near-surface observations. Often there are many unconstrained parameters such as the droplet sizes, lower tropospheric relative humidity, and rain and vapor isotopic compositions. The unique initialization and sensitivity tests that the authors do are valuable. However, it is difficult to discern the contribution of this effort to the field in its current form.
2. There is a lot of hand waving about whether the model constrained by the P3 observations can reproduce rain isotope values at the surface samples from the Ron Brown. The authors show that the model does not reproduce d-excess at the surface without a large change in relative humidity assumptions or larger drop sizes at cloud base. What do we learn about the representativeness of the model from this analysis?
3. Fig 13 is an important ‘take-away’ figure, but it is difficult to understand. The model is challenged to reproduce surface rain d-excess values of >8 permil. I’m trying to find what the starting cloud-base d-excess values are based on the P3 observations. Fig 7 is the only thing I can find and that shows model values of ~10 permil. Little change in rain drop d-excess would suggest very small rain evaporation rates. Is it possible that d-excess is not a strong indicator of evap rates? Would dD or d18O be more sensitive? Many of the figures show vertical profiles of dD or d18O, but then the model is only tested against d-excess at the surface. How does it perform against d18O and dD? What about the percentage of rain evaporated? The abstract sets up a relationship between the percent rain evaporated and d-excess, but this figure doesn’t demonstrate that link or how different the percent evaporated estimates may be in the different cases.
4. One important observation in the abstract concerns the vertical structure of rain evaporation which is sensitive to the droplet size distributions rather than the droplet concentrations. Is this droplet-resolving model unique in that regard? In other words, does the non-isotopic information provide any valuable insight as well?
5. How do the isotope observations improve understanding compared to other methods used in the field?
6. Title could be improved by mentioning the model and observations.
7. Overall, there are many figures. Are they all important for telling your story?
8. One valuable contribution that the authors could provide is making this vertically-resolved model publicly available. I encourage the authors to share their code with the community. It could be useful for providing Monte-Carlo estimates of surface rain isotopic composition in future studies.

Specific comments:

Consistent unit notation needed throughout. E.g. mm day-1, W m-2

Line 20: Which is more common in the field to describe? Evaporative flux or latent heat flux?

Line 60: ‘its’ can be unclear. Edit to mention rain drop isotopic enrichment

Line 61: define RSD at first use

Line 87: is there a citation for the reliable/unreliable size ranges?

Line 92: provide units for parameters in the equation.

Line 106-107: consider moving this sentence before “During ATOMIC”

Line 117: Mention or cite Picarro calibration and data correction.

Line 125: Several assumptions are made here. What implications does this have? In what way is the system in steady state? Equations 9 and 10 contain terms for the vapor from rain evaporation. Why state that it’s neglected?

Eqns 1 and 2: Were dD/dz and dTr/dz calculated for each diameter bin?

Line 129-130: Can you include some of the dropsonde data that confirms linear decrease in RH through the atm in a SI figure?

Line 133: Cite source of Eqn 1?

Line 134: list parameters and names one at a time so it’s easier to match up.

Eqn 3: might help if it’s shown as RWC(z).

Equations: For all equations that are evaluated at altitude (z) steps or bin sizes (i), write the equations indicating that.

Line 149: is L defined somewhere?

Line 160: was the assumption that the BL was well mixed and 150 m delta_vapor observations are representative of the BL supported by the other observations?

Line 162: All parameters obtained from Graf 2017 except the drop sizes.

Line 167: “validate the accuracy of the model” might be a reach given the current conclusions.

Line 170-174: I’m having a hard time understanding how Eqn 9 is calculated. qe is considered negligible, so qv = qva? qva is assumed constant but qv is calculated every 50 m? I’m getting stuck on what is allowed to change in the model, but doesn’t change much verses what isn’t allowed to change in the model.

Eqn 10: Earlier it was mentioned that delta_v doesn’t change with altitude?

Methods: This system of equations has parameters that feed back onto other equations. How were these solved at steady state? Iteratively? Please provide details.

Line 198: define rain frequency metric

Line 225: location of the RICO campaign?

Line 234 and Fig 5d: I do not see the negative correlation between RH and rain rate. Can you provide statistical evidence? This seems contrary to expectations.

line 236: 4 out of 5 cases were above 84%?

Line 243: ‘slightly lesser’ is awkward

Line 252 and Fig 7: I see vertical profiles in Fig 7. I don’t understand what the cases denoted by altitudes represent. Altitudes of what?

Line 327: “independently evaluating” the modeled P3 cases is misleading. There are no validation observations at the surface.

Line 335-341: The differences between Salamalikis and this study for 2 mm drops seems quite large: 64 permil vs 27 permil for dD?

3.4.1 subheading should include modeling like the 3.4.2 subheading

Line 372: give range of d-excess values rather than the spread

Line 389-390: this may be an overstatement

Line 392: remind me how the P3 case surface RH is measured? From the drop sondes?

line 412: Eqn 10 instead of 9, but the eqn doesn’t show weight.

Line 416: The conclusion that evaporated water from rain drops doesn’t influence the atmospheric vapor isotopic composition might not extend to other cases outside the tropics in drier air masses.

Fig 3: It would be more intuitive to stack the legend labels from highest altitude at the top decreasing toward the bottom.

Fig 6: boxes are difficult to see in my printed version

Fig 7: Is this modeling for the P3 case or Ron Brown case?

Fig 9: Is the red modeled or observed RSD at 700 m? While reading the description of this figure, it’s difficult to see the features that are described in the text. The relationship between droplets at 700 m and the surface are not indicated. Would arrows help? Is the log-normal fit important or can that be removed? Given the log scale, it’s difficult to identify important sizes like 700 and 900 micrometers.

Fig 11: Edit delta symbols. What RH was this model run conducted?

Fig 12: the stacked color bars do not print well. Separate histograms?

Fig A2 and others: I don’t understand the “altitude for each case.” Each case is plotted across all altitudes (0-700 m). For example, in panel d, 2 lines are shown labeled 1354 m with only 1% difference in RH, but the RWCs are extremely different. If “case altitude” and RH aren’t important, what is?

Please find my review attached.