Bayesian transdimensional inverse reconstruction of the 137Cs Fukushima-Daiichi release
- 1IRSN, PSE-SANTE, SESUC, BMCA, Fontenay-aux-Roses, France
- 2CEREA, École des Ponts and EDF R&D, Île-de-France, France
- 1IRSN, PSE-SANTE, SESUC, BMCA, Fontenay-aux-Roses, France
- 2CEREA, École des Ponts and EDF R&D, Île-de-France, France
Abstract. The accident at the Fukushima-Daiichi nuclear power plant yielded massive and rapidly varying atmospheric radionuclide releases. The assessment of these releases and of the corresponding uncertainties can be performed using inverse modelling methods that combine an atmospheric transport model with a set of observations and have proven to be very effective for this type of problem. In the case of Fukushima-Daiichi, a Bayesian inversion is particularly suitable because it allows errors to be modelled rigorously and a large amount of observations of different natures to be assimilated at the same time. More specifically, one of the major sources of uncertainty in the source assessment of the Fukushima-Daiichi releases stems from the temporal representation of the source. To obtain a well time-resolved estimate, we implement a MCMC sampling algorithm within a Bayesian framework, the Reversible-Jump MCMC, in order to retrieve the distributions of the magnitude of the Fukushima-Daiichi 137Cs source as well as its temporal discretisation. In addition, we develop Bayesian methods allowing to combine air concentration and deposition measurements, as well as to assess the spatio-temporal information of the air concentration observations in the definition of the observation error matrix. These methods are applied to the reconstruction of the posterior distributions of the magnitude and temporal evolution of the 137Cs release. They yield a source estimate between 11 and 24 March, as well as an assessment of the uncertainties associated with the observations, the model and the source estimate. The total released reconstructed activity is estimated to be between 10 and 20 PBq, although it increases when taking into account the deposition measurements. Finally, the variable discretisation of the source term yields an almost hourly profile over certain intervals of high temporal variability, signaling identifiable portions of the source term.
Joffrey Dumont Le Brazidec et al.
Status: closed
- RC1: 'Comment on gmd-2022-168', Anonymous Referee #1, 28 Sep 2022
-
RC2: 'Comment on gmd-2022-168', Anonymous Referee #2, 05 Oct 2022
Review of gmd-2022-168: Bayesian transdimensional inverse reconstruction of the 137Cs Fukushima-Daiichi release, authored by Joffrey Dumont Le Brazidec, Marc Bocquet, Olivier Saunier, and Yelva Roustan
Summary:
In this work, the authors investigated a transdimensional sampling method to reconstruct highly fluctuating radionuclide atmospheric sources and applied it to assess the 137Cs Fukushima-Daiichi release. The authors apply a reversible-jump Markov chain Monte Carlo sampling algorithm for use in Bayesian inverse problems for source reconstruction. The authors tried various methods in hopes of gaining accuracy and reducing uncertainty in the estimates, such as the inclusion of two observational sources of information (air concentration observations and deposition measurements). The authors found that the total released reconstructed activity is estimated to be between 10 and 20 PBq, which matches previous literature, although this estimate increases when considering the deposition measurements.
While the authors do a good job in explaining the methods and the results in the paper, I think there are small points that need to be addressed, mostly grammatical or in the figures.
Main comments:
This is just my opinion, but I believe that parts of the introduction that describe the Bayesian inverse modelling approach (e.g. Section 1.2) can be moved to the methods section, and parts of Section 2 describing previous literature on the topic (e.g. Section 2.1) can be moved to the introduction.
In general, the metric used to describe the goodness of fit (FAC scores) is hardly described in the text with very little background information. This makes it hard for a reader to judge the accuracy of the results. The authors can do a better job at providing some of this information for clarity. What decides if a given FAC score is considered ‘good’ or not?
The figures seem to have some mistakes or an issue with the processing. Numbers and labels drop off for many of the figures, please fix this. Also, the main text does not refer to some figures, and in some locations the reference is done with a capital letter (e.g. Figure 1) and in other locations not (e.g. figure 1). Please choose one way and keep consistent.
Figures 3-5: Is the uncertainty derived as a multiple of the standard deviation? Why not use the information from the posterior samples to apply the uncertainty, can the authors comment?
Minor Comments:
Line 7: Is MCMC defined before using here?
Line 10: Is 137Cs defined before using here?
Line 80: Take out the word ‘Here’
Lines 113-120: Various assumptions are mentioned here, but it is not clear if the impact of these assumptions on the results is discussed later in the text. It would be great if the authors can provide some comments on how these assumptions can have an influence on the results.
Line 175: A main assumption is that the average wind speed is representative of the temporally varying wind speed used in the simulations. Is this an average of the whole period? What effect does this have on the error/uncertainty of the results, compared to using a time-variable input vector for wind speed that is derived directly from observations? Again, some comments from the authors would be appreciated.
Line 204: take out the word ‘for’
Line 271: “Figure 3 shows …? “ There is a grammatical mistake here, please fix.
Line 275-285: The authors describe a ‘good match’ and ‘similar in magnitude’, but how can the reader quantify if this is true? Are the FAC scores supposed to represent this? If that is the case, the current information on the FAC score is not sufficient for the reader to make these conclusions. More information on the FAC score is needed.
Line 345: “Both the” instead of “the both”
Line 346-347: The authors state that “an adaptive grid allows to reconstruct the source term with higher accuracy and to reduce the corresponding uncertainties”. Was the higher accuracy and reduced uncertainty actually achieved? Where does the reader get this information from?
-
CEC1: 'Comment on gmd-2022-168', Astrid Kerkweg, 06 Oct 2022
Dear authors,
as a model for geoscientific model development, one important part of the journal is the publication of the algorithm implementations and models used. Please refer to https://www.geoscientific-model-development.net/policies/code_and_data_policy.html for our code and data policy.
In your article the Code availability section is missing completely. Please provide your algorithm implemenations and the models used (i.e., "the Eulerian model ldX") to the readers. They should be made available within a publicly accessible permanent archive (e.g. Zenodo).
Best regrads,
Astrid Kerkweg (GMD Executive Editor)
- AC1: 'Comment on gmd-2022-168', Joffrey Dumont Le Brazidec, 23 Nov 2022
Status: closed
- RC1: 'Comment on gmd-2022-168', Anonymous Referee #1, 28 Sep 2022
-
RC2: 'Comment on gmd-2022-168', Anonymous Referee #2, 05 Oct 2022
Review of gmd-2022-168: Bayesian transdimensional inverse reconstruction of the 137Cs Fukushima-Daiichi release, authored by Joffrey Dumont Le Brazidec, Marc Bocquet, Olivier Saunier, and Yelva Roustan
Summary:
In this work, the authors investigated a transdimensional sampling method to reconstruct highly fluctuating radionuclide atmospheric sources and applied it to assess the 137Cs Fukushima-Daiichi release. The authors apply a reversible-jump Markov chain Monte Carlo sampling algorithm for use in Bayesian inverse problems for source reconstruction. The authors tried various methods in hopes of gaining accuracy and reducing uncertainty in the estimates, such as the inclusion of two observational sources of information (air concentration observations and deposition measurements). The authors found that the total released reconstructed activity is estimated to be between 10 and 20 PBq, which matches previous literature, although this estimate increases when considering the deposition measurements.
While the authors do a good job in explaining the methods and the results in the paper, I think there are small points that need to be addressed, mostly grammatical or in the figures.
Main comments:
This is just my opinion, but I believe that parts of the introduction that describe the Bayesian inverse modelling approach (e.g. Section 1.2) can be moved to the methods section, and parts of Section 2 describing previous literature on the topic (e.g. Section 2.1) can be moved to the introduction.
In general, the metric used to describe the goodness of fit (FAC scores) is hardly described in the text with very little background information. This makes it hard for a reader to judge the accuracy of the results. The authors can do a better job at providing some of this information for clarity. What decides if a given FAC score is considered ‘good’ or not?
The figures seem to have some mistakes or an issue with the processing. Numbers and labels drop off for many of the figures, please fix this. Also, the main text does not refer to some figures, and in some locations the reference is done with a capital letter (e.g. Figure 1) and in other locations not (e.g. figure 1). Please choose one way and keep consistent.
Figures 3-5: Is the uncertainty derived as a multiple of the standard deviation? Why not use the information from the posterior samples to apply the uncertainty, can the authors comment?
Minor Comments:
Line 7: Is MCMC defined before using here?
Line 10: Is 137Cs defined before using here?
Line 80: Take out the word ‘Here’
Lines 113-120: Various assumptions are mentioned here, but it is not clear if the impact of these assumptions on the results is discussed later in the text. It would be great if the authors can provide some comments on how these assumptions can have an influence on the results.
Line 175: A main assumption is that the average wind speed is representative of the temporally varying wind speed used in the simulations. Is this an average of the whole period? What effect does this have on the error/uncertainty of the results, compared to using a time-variable input vector for wind speed that is derived directly from observations? Again, some comments from the authors would be appreciated.
Line 204: take out the word ‘for’
Line 271: “Figure 3 shows …? “ There is a grammatical mistake here, please fix.
Line 275-285: The authors describe a ‘good match’ and ‘similar in magnitude’, but how can the reader quantify if this is true? Are the FAC scores supposed to represent this? If that is the case, the current information on the FAC score is not sufficient for the reader to make these conclusions. More information on the FAC score is needed.
Line 345: “Both the” instead of “the both”
Line 346-347: The authors state that “an adaptive grid allows to reconstruct the source term with higher accuracy and to reduce the corresponding uncertainties”. Was the higher accuracy and reduced uncertainty actually achieved? Where does the reader get this information from?
-
CEC1: 'Comment on gmd-2022-168', Astrid Kerkweg, 06 Oct 2022
Dear authors,
as a model for geoscientific model development, one important part of the journal is the publication of the algorithm implementations and models used. Please refer to https://www.geoscientific-model-development.net/policies/code_and_data_policy.html for our code and data policy.
In your article the Code availability section is missing completely. Please provide your algorithm implemenations and the models used (i.e., "the Eulerian model ldX") to the readers. They should be made available within a publicly accessible permanent archive (e.g. Zenodo).
Best regrads,
Astrid Kerkweg (GMD Executive Editor)
- AC1: 'Comment on gmd-2022-168', Joffrey Dumont Le Brazidec, 23 Nov 2022
Joffrey Dumont Le Brazidec et al.
Joffrey Dumont Le Brazidec et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
418 | 83 | 12 | 513 | 4 | 5 |
- HTML: 418
- PDF: 83
- XML: 12
- Total: 513
- BibTeX: 4
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1