This study "Retrieval of ultra-violet aerosol absorption from radiation measurements in young wildfire plumes" by  Tirpitz et al. presents a novel retrieval framework to determine the imaginary refractive index (k) of biomass burning aerosols in the ultraviolet and visible spectral ranges (310-440 nm) using airborne actinic flux measurements from the FIREX-AQ campaign. The application of the VPC (VLIDORT for PhotoChemistry) radiative transfer model to dense wildfire plumes represents a technically advanced and carefully executed effort. The reported k values at 387 nm (0.02–0.03), an absorption Ångström exponent of 4 ± 1, and the identification of a bleaching half-life (~13–17 hours) provide valuable quantitative constraints on the chemical aging of brown carbon. Extending the spectral retrieval down to 310 nm is particularly impactful, as it captures the strongest absorption region of BrC. The authors also provide a transparent error propagation analysis and discuss several limitations of their inversion framework.

Overall, this is a strong and important contribution. The following comments are intended to further strengthen the physical interpretation and robustness of the retrieval.

1. The use of the 1D VPC framework is well justified for plumes that are horizontally extensive. However, the Williams Flats fire reached a maximum AOD of 11.3 at 400 nm. At such high optical depths, horizontal photon transport and 3D radiative effects can become significant even for wide plumes. The manuscript accounts for horizontal inhomogeneity through AOD perturbation within the uncertainty analysis. It would nevertheless strengthen the study to provide a quantitative sensitivity test demonstrating how the retrieved k values respond to enhanced representation of 3D effects, particularly in the densest plume regions. For example, would increasing the assumed horizontal variability or inhomogeneity term materially change the inferred k? Clarifying the expected magnitude of potential 3D-induced bias would increase confidence that the retrieval is not compensating for multi-dimensional radiative effects.
2. The retrieval assumes homogeneous spherical particles with uniform composition. Wildfire smoke, however, frequently consists of internally mixed particles (e.g., BC cores with organic coatings) and potentially non-spherical morphologies. Previous studies have demonstrated that moving from homogeneous-sphere assumptions to more physically representative mixing-state treatments (e.g., core-shell configurations) can lead to substantial differences in simulated radiative quantities (Tiwari et al., 2023). While a full reanalysis under a core-shell or fractal framework may be computationally intensive, it would be helpful to include a qualitative discussion of the expected direction and magnitude of bias introduced by the homogeneous-sphere assumption. Because particle morphology and mixing state influence not only absorption efficiency but also the scattering phase function and asymmetry parameter (g), which are essential for accurate radiative transfer calculations, it would be helpful to clarify whether part of the retrieved spectral dependence of k could reflect structural assumptions rather than purely chemical absorption changes.
3. The particle size distribution is constrained by in situ measurements and fitted to a bimodal lognormal distribution, which is commendable. However, the retrieval uses transect-averaged PSD parameters. In young wildfire plumes, rapid coagulation and condensation can modify particle sizes on relatively short timescales. If within-transect variability or systematic growth with plume age is not fully represented, some of the radiative impact of size evolution could mathematically project onto the retrieved imaginary refractive index. It would therefore be helpful to include a sensitivity analysis in which PSD parameters are perturbed within observed in situ ranges to assess the stability of k(λ). Additionally, clarifying whether PSD parameters show systematic dependence on plume age would help ensure that the inferred bleaching signal is not influenced by size evolution effects.
4. Section 4.1 notes that retrieved VAC and SSA show significantly less variability than in situ measurements, which is attributed to spatial averaging inherent in remote sensing retrievals. This is a plausible explanation. However, the discrepancy appears substantial. It would be useful to present residual diagnostics showing where the radiative transfer model deviates most strongly from the observed actinic flux. For example: Do residuals increase in regions where in situ variability is highest? Is there any systematic misfit in dense or highly heterogeneous plume segments? Such diagnostics would help clarify whether the reduced variability reflects representativeness differences or potential over-constraining of the inversion (e.g., fixed real refractive index and parameterized PSD).
5. The retrieval constrains the spectral dependence of the imaginary refractive index using a two-parameter Ångström-type power law (k₀, α_k). While this parameterization is widely used, laboratory and field studies have sometimes reported deviations from strict power-law behavior in the near-UV, particularly below ~350 nm. Could the authors comment on the sensitivity of the retrieved α_k and bleaching timescale to this spectral constraint? A brief discussion or limited sensitivity test allowing small deviations from the power-law form would help clarify the robustness of the inferred aging signal.
6. Table 1 indicates differences in fuel type and combustion characteristics across the sampled fires. The manuscript reports similar bleaching timescales (13–17 hours) and notes no statistically significant differences among fires. It would be interesting to show fire-specific half-life fits (with uncertainty ranges) to clarify whether the apparent similarity reflects true physical consistency or limited statistics. A short discussion on whether fuel type, combustion efficiency, or initial chemical composition might influence bleaching rates would enhance the interpretation.
7. A brief discussion of how vertical variability observed during aircraft sampling is represented (or averaged) in the 1D radiative transfer framework would further clarify the robustness of the inversion.

Reference:

Tiwari, P., Cohen, J.B., Wang, X. et al. Radiative forcing bias calculation based on COSMO (Core-Shell Mie model Optimization) and AERONET data. npj Clim Atmos Sci 6, 193 (2023). https://doi.org/10.1038/s41612-023-00520-1