Review of the manuscript “ Characterisation of uncertainties in an ocean radiative transfer model for the Black Sea through ensemble simulations” by Loïc Macé et al.

The authors investigate the influence of uncertainties in the inherent optical properties on the modelling of radiometric quantities adopting a radiative transfer model applied to the Black Sea. A detailed study is presented that considers several important optical components and discusses their role in light propagation. An advanced stochastic approach is used to understand the uncertainty and the relative contribution of plankton, nap, CDOM.

The results are clearly explained, the approach used is new and represents an important advance for biogeochemical modelling as it increases the realism of the models. Moreover, the analysis  presented could be useful for the marine biogeochemical modelling community.

Below I list some minor revisions that I think could improve the text.

Pg2 lines 47-48 “Over the years, various models were developed with different approaches to refine the representation of light in models (Gregg and Casey, 2009; Mobley et al., 2009).” I would remove the repetition of models.

Pg 3 lines 69-70 “The penetration of the spectral irradiance is determined by the absorption and scattering properties of the medium that are derived from concentrations of optically active components, in 33 wavelengths.” I would substitute with “The penetration of the spectral irradiance is determined in 33 wavelengths by the absorption and scattering properties of the medium that are derived from concentrations of optically active components.”

Pg 6 Eq.4 I would give the unit of measurement of PAR. Since the previous section mentions the one-way coupling with the RT model, I would explain why the PAR is introduced by the RT model since there is no feedback with biogeochemistry.

Pg 15 Ensemble simulations. I would suggest that the authors create a table summarizing the properties of each of the 4 experiments.

Pg 17 line 426. “until in increases” should be “until it increases”.

Pg 18 Figure 6 is very small and very difficult to read. Units of measurement in the y-axis are missing. In general, many of the illustrations in this manuscript are very small, especially the fonts, and the fonts, size and legibility should be enlarged.

Pg 22 Figure 10,  again the figure is very small. PAR should normally be expressed in quanta, and the PAR data from the BGC Argo float is also normally expressed in quanta. How did you convert to have W/m2? Perhaps the conversion formula could explain the discrepancy in PAR in Figure 10?

Pg 22 lines 513- 521 Rank diagrams should be explained in more detail so that they are fully understandable, especially for the readers without expertise in ensemble modelling.

Pg 22 lin 524 iis the reference to Fig. 9 correct? The matching between  model and data in Fig. 9 are very good, so the comment is unclear.

Review of “Characterisation of uncertainties in an ocean radiative transfer model for the Black Sea through ensemble simulations” by Loïc Macé et al.

The authors use a biogeochemical ocean model coupled one-way to a 3 stream radiative transfer model to show how modelled chlorophyll based on modelled spectral reflectance band ratios are more consistent with both satellite ocean colour and biogeochemical Argo float data than chlorophyll estimates derived directly from their physical-biogeochemical coupled model system, NEMO-BAMHBI. They demonstrate that introducing uncertainties in the form of random perturbations of inherent optical properties of different water constituents, improves the simulated distributions of radiative fields. The study is focused only in the deep central areas of the Black Sea, due to the limitations of the CDOM forcing used in the experiment.

This is an excellent piece of work. It significantly advances our understanding of how inherent optical properties of water constituents can be used to constrain biogeochemical models, leading to improved modelled predictions. Moreover, it opens new possibilities for integrating biogeochemical modelling, in situ optical observations and ocean colour remote sensing. The authors also demonstrate that preliminary results from a two-way coupling test show even more promising results.

The paper is well organised and well thought out. The approach and methods are valid. Limitations and assumptions inherent in their approach are discussed in depth and perspectives for future work are provided. I would be happy to see this published in Biogeosciences. It is a very timely and welcome contribution which unifies in situ marine optics, satellite ocean colour and biogeochemical modelling communities.

I have listed some minor comments below for the authors to consider.

1. All of the figures (except figure 1) should be bigger, with larger fonts on the labelling.
2. A table summarizing the different reference and ensemble runs would be useful.
3. Some references are missing, or in the wrong place.
   • Line 36: add Cahill et al., 2008, https://doi.org/10.1029/2008GL033595
   • Line 49: add Bissett et al., 1999, https://doi.org/10.1016/S0967-0637(98)00063-6
   • Dutkiewicz et al., 2015 should be referenced earlier, I think, line 59 after the sentence “In recent years, … reflectance (Dutkiewicz et al., 2015)
4. In places, the English should be improved for better understanding, e.g.
   • Line 39: “They can be described by the absorption and (back-) scattering spectra of each water …” instead of “They consist in absorption and …”
   • Lines 67-68: at the end of the sentence “It solves the spectral wavebands corresponding to those typically used in remote sensing” add an example of what these wavebands are?
   • Lines 86 – 89: Suggest converting the following into a numbered list, easier to comprehend sequence of analysis. “We first assess the effect of … that are consistent with observations.”
   • Line 96: change last sentence to something like: “Finally in section 5, we discuss the limitations and assumptions of the study and provide an outlook for future work.”
   • Line 109: change to “ … solved with NEMO 4.2 which is online coupled to the biogeochemical model.”
   • Line 143: change to “Attenuation coefficients are derived from absorption …”.
   • Lines 169 – 172: this paragraph appears without any explanation, should be qualified with a statement which explains why the product is mentioned, e.g. to validate the modelled reflectances.
   • Lines 179-181: explain more clearly ecological reasons for choosing the band-ratio algorithm over the NN approach. Your study is focused on waters where the reflectance signal is dominated by phytoplankton, for example?
   • Lines 207 – 208: change to “ … solved in BAMHBI: these are large flagellates ….and diatoms, all of which are the dominant species in the Black Sea”
   • Line 249: change to “When run over the Black Sea, ….”
   • Line 260: change to “… water column, the water constituent IOPs would not be altered.”
   • Lines 365 – 371: add table summarizing different simulations.
   • Line 426: change to “ … until it increases again …”
   • Line 430: change to “… at 555 nm, as its contribution to …”
   • Lines 447 - 448: change to “… phytoplankton early in the year with a lower contribution of CDOM in the ensemble spread.”
   • Line 457: change to “ … gradually increases with depth …”
   • Line 562: “surface chlorophyll is defined as the average concentration over the top 10m.” Is this based on some average of the 1^st optical depth? Or? Maybe elaborate a little on this choice of depth over which to integrate the data.