the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A synthetic database generated by radiative transfer simulations in support of studies in ocean optics and optical remote sensing of the global ocean
Hubert Loisel
Daniel Schaffer Ferreira Jorge
Rick A. Reynolds
Dariusz Stramski
Abstract. Radiative transfer (RT) simulations have long been used to study relationships between the inherent optical properties (IOPs) of seawater and light fields within and leaving the ocean from which the ocean apparent optical properties (AOPs) can be calculated. For example, inverse models to estimate IOPs from ocean color radiometric measurements have been developed or validated using results of RT simulations. Here we describe the development of a new synthetic optical database based on hyperspectral RT simulations across the spectral range from the near-ultraviolet to near-infrared performed with the HydroLight radiative transfer code. The key component of this development was the generation of the synthetic dataset of seawater IOPs which served as input to RT simulations. Compared to similar developments of optical databases in the past, the present dataset of IOPs is characterized by probability distributions of IOPs that are consistent with global distributions representative of vast areas of open ocean pelagic environments and coastal regions covering a broad range of optical water types. The generation of the synthetic data of IOPs associated with particulate and dissolved constituents of seawater was driven largely by an extensive set of field measurements of the phytoplankton absorption coefficient collected in diverse oceanic environments. Overall, the synthetic IOP dataset consists of 3320 combinations of IOPs. Additionally, the pure seawater IOPs were assumed following recent recommendations. The RT simulations were performed using 3320 combinations of input IOPs assuming vertical homogeneity within an infinitely deep ocean. These input IOPs were used in three simulation scenarios associated with assumptions about inelastic radiative processes in the water column and three simulation scenarios associated with sun zenith angle. Specifically, the simulations were made assuming no inelastic processes, the presence of Raman scattering by water molecules, and the presence of both Raman scattering and fluorescence of chlorophyll-a pigment. Fluorescence of colored dissolved organic matter was omitted from all simulations. For each of these three simulation scenarios, the simulations were made for three sun zenith angles of 0°, 30, and 60° assuming clear skies, standard atmosphere, and wind speed of 5 m s−1. Thus, overall 29880 RT simulations were performed. The output results of these simulations include the radiance distributions, plane and scalar irradiances, and the whole set of AOPs including the remote-sensing reflectance, vertical diffuse attenuation coefficients, and mean cosines where all optical variables are reported in the spectral range from 350 to 750 nm at 5 nm intervals for different depths between the sea surface and 50 m. The consistency of this new synthetic database has been assessed through comparisons with in situ data and previously developed empirical relationships involving the IOPs and AOPs.
Hubert Loisel et al.
Status: open (until 06 May 2023)
-
RC1: 'Review of the manuscript “A synthetic database generated by radiative transfer simulations in support of studies in ocean optics and optical remote sensing of the global ocean”, by Loisel et al.', Jaime Pitarch, 15 Mar 2023
reply
-
RC2: 'Comment on essd-2023-80', Anonymous Referee #2, 27 Mar 2023
reply
General appraisal
The manuscript by Loisel and co-workers presents a synthetic data set of so-called apparent optical properties (AOPs) generated through simulation of the radiative transfer (RT) in oceanic waters. The goal is to help the development and test of inversion algorithms that use these AOPs as input (generally the remote-sensing reflectance, Rrs) and attempt to derive quantities such as the inherent optical properties (IOPs, here the absorption and backscattering coefficients) or biogeochemical / ecological properties such as the phytoplankton chlorophyll concentration (Chl). The Rrs are generally derived from satellite ocean colour observations, although the inversion can also be applied to Rrs derived from in-situ radiometry measurements.
Numerous such data sets exist and the justification for proposing this new one is that the IOPs used as inputs to the RT simulations are made more representative of the real world by using probability distribution functions that are consistent with what global in-situ or satellite data sets reveal.
I guess this is an important feature if the data set is then used in its entirety to, e.g., train ML-based algorithms such as neural networks. In such a case, it is indeed critical that the training data set is as realistic as possible. If the data set is used by expert users who can appropriately select what they need in this simulated data (e.g., they can pick up whatever range of the various input parameters that they think is relevant to their development), then the justification for this new data set is a bit weaker.
In any case, I think it is a very useful data set. I appreciate that the authors have included not only the Rrs spectra but also most apparent optical properties as well as the full radiance distributions. This will undoubtedly make this dataset of a broader interest.
The effort to make the synthetic data set better representative of the real global ocean, which is largely made of clear and moderately clear waters, is also a very good one. As clearly shown by Fig. 6, previous synthetic data sets did not match at all what the global ocean is. They were largely skewed towards high values of most IOPs, which is rather typical of coastal waters. Still, they were often used to develop and test algorithms that are then applied globally. To my opinion, this situation has led to frequent overstatement about the performance of inversion algorithms applied to satellite ocean colour.
The use of the Hydrolight RT code is a relevant choice, in particular because it allows Raman scattering and Chl fluorescence to be simulated. I would however recommend that the authors make clearer in the paper that modelling Chl fluorescence includes several assumptions so that any algorithm development using the part of the spectrum affected by Chl fluorescence will be largely depending on those assumptions. They allude to this on lines 577-578 (by mentioning that the quantum efficiency of Chl fluorescence was set to a default value) but a better description of this part and of the associated limitations should be included.
The selection of possibly realistic combinations of absorption by phytoplankton, coloured dissolved organic matter (CDOM) and non-algal particles is important if users of the data set aim at retrieving these quantities. However, the authors know that RT in the ocean, for given boundary conditions, is entirely driven by only two inherent optical properties: the total absorption coefficient and the total volume scattering function (VSF). It does not matter for RT whether a given total absorption at a given wavelength is generated by various proportions of phytoplankton or water or anything else. The important parameters are therefore the ratio of particulate to molecular scattering (defining then the total VSF) and the ratio of total absorption to total scattering or total attenuation. Therefore, although I recognise that it is useful to see IOP distributions for components like phytoplankton and CDOM (Fig. 2), it would be equally useful to see the distributions of the parameters mentioned above. With what is presented, we cannot be sure that they are well covered.
A few more detailed comments
- Title: a data base of what? Maybe that could be said (Although the title is already quite long)
- The entire section 2 is quite dense and not that easy to read. Some sub-headings might help better figure out the various steps followed in generating the IOP set. I guess many details could also be moved into a supplemental section, at least if the Journal allows it.
- Line 115: the use of Zhang et al to calculate the seawater scattering coefficient implies that a temperature and a salinity have been chosen. This should be indicated, and it should be made clear that this data set cannot be used for freshwater, for instance. The manuscript title says “global ocean” but we know that not all users might be that careful.
- Paragraph 199-214: the in-situ data bases that have been used should be listed and acknowledged. I know it can be a bit cumbersome because there are many, but a Table would make this efficiently. Citing papers that have previously used these data bases is not the correct way of doing it.
- The spectral values of the particulate backscattering coefficient resulting from summing up the phytoplankton and non-algal particle backscattering coefficients derived through Eqs. 8 and 11 should be displayed for some Chl values (or the spectral slope as a function of Chl could be displayed). That is an important parameter and it is unclear how it looks like.
- Pages 12-13: not sure I have got the rationale for the P1, 2, 3, 4 parameters and corresponding equations in Table 1. This should be better explained.
- Pages 5-9: this is quite long for describing how absorption is modelled. And, there is conversely little about scattering (discussing the phase function only comes at page 14).
- Fig. 9 would be advantageously completed by a similar plot of Rrs490/Rrs555 vs. Chl.
- Another very important parameter to show would be the ratio of CDOM absorption to total non-water absorption vs. Chl. I guess what I am trying to say here is that a more comprehensive assessment of how this data set compares to previous ones and to established Case 1 water bio-optical models would be really useful.
Citation: https://doi.org/10.5194/essd-2023-80-RC2
Hubert Loisel et al.
Data sets
A synthetic database of hyperspectral ocean optical properties H. Loisel, D. S. F. Jorge, R. A. Reynolds, and D. Stramski https://datadryad.org/stash/share/2OnzP-2wlb3jYrQtjjIp7_DmYoJfNK7aTGqZWhqYuhw
Hubert Loisel et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
269 | 59 | 10 | 338 | 2 | 3 |
- HTML: 269
- PDF: 59
- XML: 10
- Total: 338
- BibTeX: 2
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1