The manuscript presents a data-based method for estimating ocean anthropogenic carbon dioxide concentrations. The method can be used to estimate as a function of location, temperature and salinity the geographical patterns for both historical and projected future atmosphere CO2 levels. The "tracer-based rapid anthropogenic carbon estimation" (TRACE) method uses recent compilations of ocean transient tracer field measurements (e.g., SF6 and CFCs) to compute water parcel age distributions (transient time distribtions or TTDs) as a function of location and water mass properties. Anthropogenic CO2 perturbations to the dissolved inorganic carbon field are then calculated from the age distributions using atmosphere CO2 time histories and assumptions about air-sea dis-equilibrium. Neural network methods are used to approximate the required intermediate steps and final anthropogenic CO2 product across the full range of geographic locations and water mass conditions. The method is tested against various model and observation-based anthropogenic CO2 estimates. The manuscript is through and well written, and it includes sufficient depth on the methods, assumptions, and uncertainties. The TRACE method builds on a couple of decades of research connecting ocean transient tracers to anthropogenic carbon and will likely be beneficial to a range of ocean scientists and other users.

Overall the manuscript is a solid scientific contribution, and I only have two issues that I recommend the authors consider addressing. In both cases, I am not suggesting that the authors modify the methodology, simply that they include a little more detail on caveats that may be useful for some readers.

In Line 153 in "Section 3.2 How TRACEv1 work" the text states: "1. uses a neural network to estimate an age distribution for seawater from a user-specified location, T, and S, and returns the mean age if this is a desired output;"

It would be helpful to add some text discussing the rationale for the variables chosen for the neural network. In particular, for many potential users location and depth would be a more straight forward variable set than location, T, and S. The need for T and S are clear for CO2 solubility but not all potential users have knowledge of T and S distributions. Presumably the neural networks are connecting T and S to density to approximate depths of isopycnal surfaces, recognizing that TTDs reflect advection and mixing along isopycnal surfaces as the primary path for the introduction of anthropogenic CO2 into the ocean interior, at least in the thermocline.

On a related issue, T and S fields in the ocean have already changed with time and will evolve further with future climate change; warming of the thermocline (and shoaling of isopycnal surfaces) is already well documented, altering ocean ventilation rates, and these trends will continue in the future, particular under strong climate change projections. The model uses time-invariant T and S fields, thus neglecting changes in ocean properties (including CO2 solubility) and ventilation rates. To incorporate these secondary effects, users could consult climate model projections that include time-evolving ventilation patterns, transient tracer and the projected anthropogenic CO2 fields. There is a caveat in the manuscript to this effect on line 430:

"An important caveat is that these findings do not consider the impacts of changes in heat and freshwater content, circulation, or changes in the ocean’s biological pump, and only reflect the impact expected from changing atmospheric xCO2 and ocean buffer capacity." and then again on Line 454: "It presumes fixed circulation and is unable to resolve most timescales and modes of Canth variability." However, it would be beneficial to explicitly include this caveat in the Abstract so readers no up front potential limitations (or alternatively, areas for future enhancement).

Review of Tracer-based Rapid Anthropogenic Carbon Estimation (TRACE) by Carter et al.

This study use presents a tool that has been developed to estimate the storage of anthropogenic carbon in the interior ocean using observational and model data. It is a valuable contribution to the science on the ocean carbon uptake. The question – how much of the observed DIC in the interior ocean is in excess to what was there before the industrial revolution – is a difficult one, and the authors identify (at least) 3 main hurdles.

The manuscript present 3 products: 1) The TRACEv1 code, 2) uncertainty estimates, and 3) a global data product.

The paper is well written, timely, and of potential high impact, and deserves to be published. I do however see a few areas where I would like the authors to improve on the manuscript.

It is unclear how the variable alpha (that reduce the TTD shape to a single term) is calculated – please provide this information in the manuscript.

Line 60: This sentence is difficult to understand. Please rephrase.