the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Bayesian assessment of chlorofluorocarbon (CFC), hydrochlorofluorocarbon (HCFC) and halon banks suggest large reservoirs still present in old equipment
- 1Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
- 2NOAA Chemical Sciences Laboratory (CSL), Boulder, CO 80305-3328, USA
- 3NASA Goddard Space Flight Center, Greenbelt, MD, USA
- 4Science Systems and Applications, Inc., Lanham, MD, USA
- 5Laboratory for Air Pollution/Environmental Technology, Empa, Swiss Federal Laboratories for Materials Science and Technologies, Duebendorf, Switzerland
- 1Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
- 2NOAA Chemical Sciences Laboratory (CSL), Boulder, CO 80305-3328, USA
- 3NASA Goddard Space Flight Center, Greenbelt, MD, USA
- 4Science Systems and Applications, Inc., Lanham, MD, USA
- 5Laboratory for Air Pollution/Environmental Technology, Empa, Swiss Federal Laboratories for Materials Science and Technologies, Duebendorf, Switzerland
Abstract. Halocarbons contained in equipment such as air conditioners, fire extinguishers, and foams continue to be emitted after production has ceased. These ‘banks’ within equipment and applications are thus potential sources of future emissions, and must be carefully accounted for in order to evaluate ongoing compliance with the Montreal Protocol. Here, we build on a probabilistic Bayesian model, previously developed to quantify CFC-11, 12 and 113 banks and their emissions. We extend this model to the suite of the major banked chemicals regulated under the Montreal Protocol (HCFC-22, HCFC-141b, and HCFC-142b, halon-1211, and halon-1301, and CFC-114 and CFC-115) along with CFC-11, 12 and 113 in order to quantify a fuller range of ozone-depleting substance banks by chemical and equipment type. We show that if atmospheric lifetime and prior assumptions are accurate, banks are very likely larger than previous international assessments suggest, and that production has been very likely higher than reported. We identify that banks of greatest climate-relevance, as determined by global warming potential weighting, are largely concentrated in CFC-11 foams and CFC-12 and HCFC-22 non-hermetic refrigeration. Halons, CFC-11, and 12 banks dominate the banks weighted by ozone depletion potential. Thus, we identify and quantify the uncertainties in substantial banks whose future emissions will contribute to future global warming and delay ozone hole recovery if left unrecovered.
- Preprint
(571 KB) -
Supplement
(258 KB) - BibTeX
- EndNote
Megan Jeramaz Lickley et al.
Status: final response (author comments only)
-
RC1: 'Comment on acp-2022-240', Anonymous Referee #1, 25 Apr 2022
Review for “Bayesian assessment of chlorofluorocarbon (CFC), hydrochlorofluorocarbon (HCFC) and 2 halon banks suggest large reservoirs still present in old equipment” by Lickley et al.
This manuscript presents an extension to earlier work, where ODS banks, and their emissions, are estimated in a statistical framework using uncertain knowledge about emission profiles and atmospheric mole fraction measurements. This study is timely, clear and concise, and will be of broad interest to readers of ACP. I have only a few suggestions, with the majority concerning clarifications in the text and some additional discussion. I have one concern over the treatment of what is termed ‘reported production’ and how this may impact the conclusions drawn, which I outline below. I hope to see the eventual acceptance and publication of this manuscript once my comments have been addressed.
Comments:
- Production-related emissions and leakage rates: From my understanding, reported production is the net production, i.e. reported after any losses during the production process, or the “sellable production”. As such, at least in theory, cumulative production should equal consumption for CFCs. This means that total production, before any production losses, would be larger than that reported, i.e. Total production = P/(1-DE) using the notation in the paper (where P is reported production), and production-related emissions would be equal to DE*P/(1-DE). What I take from the manuscript is that, currently, the production-related emissions are quantified instead from the production that has been reported after the losses have occurred. I don’t think this will alter the conclusions of the paper, but it may close some of the gap and reduce some of the reported bias, as it may mean a few percent more production is added to the bank.
- Discussion of discrepancies in production: Non-dispersive (and therefore not required to be reported) production exists, for example when gases are produced as by-products. An example CFC-115 contamination in the production of HFC-125 (see Vollmer et al 2018). I don’t believe this is a reason for the discrepancies, and there’s no evidence to support or deny this, but should be discussed in addition to dispersive and feedstock related production.
- End-of-life emissions: The impact of end-of-life emissions needs further discussion. How would a change in the emissions rate due to disposal change the conclusions drawn? Some information exists surrounding post-disposal emissions, e.g. from the US EPA (see e.g. page A-262, Table A-127 of https://www.epa.gov/system/files/documents/2022-04/us-ghg-inventory-2022-annex-3-additional-source-or-sink-categories-part-a.pdf). I imagine this would be too difficult to include in any analysis (and I’m not sure if there is sufficient information to do this) but there is a need to expand the discussion around long-lived banks.
Technical comments:
Abstract, line 19: “…must be carefully accounted for in order to evaluate ongoing compliance with the Montreal Protocol”. This could be interpreted that these future emissions fall under controls of the Montreal Protocol, even though emissions from the bank are not controlled. Perhaps better is suggest the importance is to evaluate nascent production vs. banked emissions, or similarly you could change the ‘impact’ to be for something like stratospheric ozone recovery.
Abstract, line 22: “model a suite” rather than “the suite”
Abstract, line 27: “production for dispersive uses” or similar, as production will be higher than that reported due to e.g. by-product emissions and point of production losses.
Abstract, line 32: Delay ozone recovery in reference to what? Current projection generally take banks into account.
Line 38: *stratospheric* ozone depletion
Line 48: On the use of “we developed…”. “we” here does not envelop all co-authors, so perhaps better to refer to it either in the passive, or “Lickley et al 2020, 2021 developed…”
Line 55-57: It would be clearer to continue to use the term ‘reported production’ rather than just ‘production’ here.
Line 75: Are these cumulative emissions derived using a top-down or bottom-up method?
Line 86: It’s generally called “Bayes’ theorem” rather than “theory”
Line 127, eq. 3,4: Subscript italics should be saved for variables, i.e. ‘Total’ should not be italic.
Line 141, eq 5: It is not clear where the constants A come from. Is it from the SPARC 2013 reference? If not they should be included in the manuscript.
Line 149: Does the term ‘fire extinguishers’ here refer to all forms of fire and explosive suppressant? Halons are used in many other applications that only fire extinguishers. If this is only a general term used by AFEAS then this should be made clearer.
Line 157: It’s not clear what functional form the correlation takes (this is also not clear in the supplement), only the prior set on the correlation parameter is specified. I.e. is there an assumed linear covariance? Or something like autocorrelation?
Line 173: What’s a ‘medium value’? Should this be median?
Line 189: There should be a full stop instead of a comma before ‘therefore’
Line 196: NOAA needs defining
Line 422: The reference to Vollmer et al. 2018 has a formatting error
Table 2: It would be very useful to also include the absolute difference in the estimates (with uncertainties) here in Gg, in additional to only the percentages, to get a better sense of the importance of these various compounds.
Figure 3: Is there a significance to the circle size in Fig 3? If this is to distinguish between WMO 2018 and TEAP 2009 only then perhaps choose a different shape or line style.
-
RC2: 'Comment on acp-2022-240', Anonymous Referee #2, 06 May 2022
This is a nice extension of a previous analysis to include more chemicals and more processes affecting past and potentially future emissions of halocarbons from banks. The work is important and highly relevant to current issues about halocarbons and Montreal Protocol compliance. I had 2 main issues that need some discussion and exploration before the paper is publishable.
Some consideration of end-of-life processes (sensitivity or includes as a separate category) is needed for banks for which end-of-life emission might be substantially different from emission rates during use (particularly cc foam for CFC-11, perhaps also non-hermetic refrigeration for CFC-12 and HCFC-22). TEAP reports have suggested that this could be a significant influence on emissions in recent and future years, yet this process is not considered by the authors owing to their view of a lack of information (line 329). I’d suggest that some exploration or sensitivity analysis of the issue is important to increase the relevance of these results. The Bayesian approach provides optimized parameters for the past, and those parameters may not be relevant for the future when the relative contributions of end-of-life emission increases substantially. For the model to provide useful expectations of emissions in the future, it must accurately represent a future where emissions are dominated by processes not as prominent in the past, i.e., end-of-life.
It isn’t clear if the AGAGE mole fractions being fit in this analysis are surface means or some representation of total atmosphere average mole fractions. I suspect that observation-based surface mean mole fractions are being used 'as is' to represent total atmosphere mean abundances and, if so, some further consideration is needed. For nearly all of these gases the bias between these two quantities was substantial in the past (up to 20% theoretically but likely less for most years), varies over time (reduced in recent years as emissions are less), and might add substantially to the larger than reported production and estimated bank sizes argued for in the present analysis. Related to this point (vertical mole fraction gradients are substantial and time-varying) a more realistic and time-varying relationship between mole fraction and emission (equation 5) needs including if indeed surface mole fractions are what is included in the inversion analysis.
Details:
The abstract makes assertions that seem too strong given the substantial caveats mentioned at the very end (lines 325-334). These caveats seem outside the assumptions related to priors and lifetime that are mentioned or even hinted at in the abstract.
Line 69, accounting methods use reported information, but also estimates and assumptions about processes leading to emissions.
Figures 1 and 2 show observationally-based results and emissions for only a fraction of years for a number of gases. Inversions would seem to be less relevant and less accurate if performed on these limited data histories.
Citations of Assessment reports should be called out by the lead authors’ names in most if not all places (not WMO 2018 or SPARC, 2013). It is only done the accepted way in a few instances in the manuscript.
Line 321-3. Are these really the only two possible explanations for 113?
-
RC3: 'Comment on acp-2022-240', Anonymous Referee #3, 08 May 2022
This well-written manuscript uses a probabilistic Bayesian model to quantify residual storage (banks) of multiple ozone-depleting substances that are released to the atmosphere even after their production has been curtailed by regulation. The method appears to be a valuable approach to checking compliance to the Montreal Protocol for multiple compounds, and earlier iterations of this method have been used to infer lifetimes and banks of CFC-11, CFC-12 and CFC-113. From my reading, it appears to mesh fairly well with observation-based approaches that use background concentrations, global transport models, and inverse modeling to derive emissions estimates (e.g., the concurrently submitted paper to this journal on HCFC-142b by Western et al). I am not intimately familiar with either modeling approaches, so my goal here is to enhance the readability of the present manuscript to expand accessibility to larger audiences. The following are suggested as items for clarification.
- Introducing the terms “prior distributions” and “priors”. It would be helpful to define these terms to help readers who are not familiar with such terminology. Lines 145-147 could be clarified as follows: The input parameters in the simulation model described above require initial values to be assigned, along with their probability distributions. These prior distributions (‘priors’) are developed to estimate mole fractions, emissions, and banks for CFC-11, 12, 113, 114, and 115, HCFC-22, 141b, and 142b, and halon-1201, and 1311.
- Lines 187-190. This sentence has a grammatical issue. “While there are published estimates of uncertainties in observed mole fractions, the uncertainties in modeled mole fractions do not, therefore, we estimate S separately for each chemical…”
- Table 3 conversions. I’m not sure how to interpret the units for GWP100 and OPD. For the GWP100, is it Gg CO2 equivalent per year? For ODP, is it Gg CFC-11 equivalent per year?
- Disparities in CFC-115. Because the Bayesian model differs from the observed CFC-115 mole fractions, and the modeled emissions are very different from the observationally-derived emissions, how much confidence should we have regarding the magnitude of the bank estimates or emissions by source for this compound?
- Unexpected differences between Figures 3 and 4. For most compounds, I would expect the emission rate by source (Gg/yr, Fig 4) to be a fraction of the magnitude of the banks (Gg, Fig 3). This is the case for CFC-11, CFC-12, HCFC-22, F141b, and F142b, all of which appear to have an emission rate of ~10-20% of the banks per year. However, CFC-113 appears to have an emission rate that exceeds the entire bank size per year, CFC-114 appears to be 3 orders of magnitude larger, and F-115 appears to be 2 orders of magnitude larger. Is that correct? If so, does that mean that essentially all of those compounds are dispersed immediately (i.e., that banks are inconsequential)? Then why are CFC-115 emissions coming entirely from long-banks? I seem to be missing something important here.
- Different time ranges on x-scale. This is a minor edit, but it would help the reader line up plots if the x-axes for the figures used the same time range. Plots start in 1940, 1950, mid 1950s and 1960.
- In order to contextualize these results with prior studies, it might also be helpful to include the results from Lickley et al, 2020 and Lickley et al. 2021 on some of the plots.
Overall, this is a useful extension of their prior studies, and the study will shine a light on the discrepancies between what is reported and what is happening in terms of these regulated halocarbons. Given the importance, I think the above clarifications will help make the manuscript more accessible to a broader audience.
Megan Jeramaz Lickley et al.
Megan Jeramaz Lickley et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 273 | 50 | 12 | 335 | 19 | 1 | 2 |
- HTML: 273
- PDF: 50
- XML: 12
- Total: 335
- Supplement: 19
- BibTeX: 1
- EndNote: 2
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1