the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 May 2026
Addressing systemic underestimation in global ship emissions from fleet growth and fuel compliance
Abstract. As a hard-to-abate sector, global shipping is under international and regional emission control regulations. To evaluate emission control effects and conduct rapid response air quality simulations, accurate and timely ship emission inventories are indispensable. However, current ship emission inventory models face multiple challenges, including incomplete and delayed global ship fleet description and significant divergence in PM2.5 emissions after global low sulfur regulation came into effect. Here, we established a dynamically updated Ship Emission Inventory Model that allows near-real-time emission calculation. Ship activity and technical database are updated daily instead of yearly to obtain a more complete and rapid description of global ship fleet. Fleet's multiple choices to comply with fuel sulfur regulation were considered, including switching to very low sulfur fuel and utilizing scrubbers to keep consuming heavy fuel oil. The daily expansion of ship technical database uncovered 8% and 6.2% of the total gross tonnage of active bulk carrier and container fleets, unveiling up to 5.4% of global ship CO2 emissions. Without the expansion, the daily underestimation would enlarge over time from about 0.20 Mt CO2/d to 0.29 Mt CO2/d throughout 2024. On the other hand, the single compliance choice assumption and ignorance of heavy fuel oil use after 2020 would lead to underestimation of PM2.5 and BC emissions by approximately 55% and 27%. Although South China Ocean had the most absolute underestimation, the Indian Ocean had the highest underestimated portion, reaching 75% and 39% of its total PM2.5 and BC emissions.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(4717 KB) - Metadata XML
-
Supplement
(1156 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-1935', Anonymous Referee #1, 12 Jun 2026
- AC1: 'Reply on RC1', Huan Liu, 06 Jul 2026
-
RC2: 'Comment on egusphere-2026-1935', Anonymous Referee #3, 20 Jul 2026
Review of the paper “Addressing systemic underestimation in global ship emissions from
fleet growth and fuel compliance” by Weiwei Zhang and co-authors.
The paper describes model updates to the Shipping Emission Inventory Model and their consequences for global ship emission data. The authors give the impression that their findings are quite general and not specific to their own model development only. I think this paper would be much better suited for a model development journal like Geoscientific Model Development than for Atmospheric Chemistry and Physics. In any case, it is important that the authors make clear from the beginning that they mainly speak about correcting deficiencies in their own model system rather than general findings about global shipping emission calculations. Also, the so-called “single compliance choice assumption” for selecting a technology to comply with global sulfur limits is a simplification that is potentially highly inaccurate and should not be used if better information is available from a ship technical database.
The paper also needs major improvements in the English language and a more precise description of the findings. A real discussion of the uncertainties would also help to make the paper more relevant for a wider audience.
Major specific comments
Line 27: You need to specify which species you refer to. The number is different for CO2, NOx, SO2 … .
Line 31: The number of 17% is highly dependent on the global development of CO2-emissions in all non-shipping sectors. You need to mention this.
Lines 34-37: You need to say between which years these reductions took place, i.e. when the policies were implemented.
Line 42/43 and line 45: Which ship emission model(s) are meant here? It seems you speak about ship emission models in general, but you do not mention any.
Line 50: It is unclear how all of these numbers go together. When 65% of the fuel sales are (high sulfur) HFO, there must be many ships being equipped with scrubbers. What is the number of scrubber ships in your model system? It must be much higher than 800. And how does the number of 65 % HFO agree with Fig. S4, which shows lower values for HFO for three ship size classes?
Line 56: What are “newly build ships” in your terminology? Are these ships younger than 5 years old? Why should their number increase over time?
Line 66/67: Why wasn’t this done before? And: Is the technical information included in the AIS data detailed enough to improve the emission model? Typically, the AIS data is not as detailed as that in a ship characteristics database.
Line 77: It is unclear what you mean by AIS data that is updated daily instead of yearly. Isn’t the AIS data always historic, i.e. collected over a longer time and then used as input for SEIM?
Line 81: You should explain how ships were identified in SEIM v2.2? Is there a ship characteristics database that represents the fleet composition in a certain year? Why didn’t you use the ship information contained in the AIS data before?
Line 98: Please explain where or from whom you get the AIS data.
Line 112: When you need to perform a three-day AIS processing to get reliable data, how does this agree with your “near-real-time” approach?
Line 128: When I understand this correctly, STSD contains more information, e.g. on engine power, compared to AIS data. How do you solve this problem?
Fig. 2: Why are there only 5 ship types? In line 131 you say that SEIM has 14 ship types. And why are there more than 400,000 ships in the database when there are only approx. 100,000 being active (see line 82)? Why is there a sharp increase in the number of identified new ships between July and October 2024?
Line 90: What is meant by “well cleaned”? How much data is omitted during the cleaning process?
Line 142: What you call here the “information of [a] newly detected ship” is a guess (or interpolation) based on existing information. Similar methods were already applied before in models like MoSES (Schwarzkopf et al., 2021) and STEAM (Jalkanen et al. 2009,2012).
Line 150-153: BC is not included in Table S3, so what are the reductions based on? Is BC taken as a constant fraction of PM2.5? Is this a correct assumption? BC emissions should depend strongly on combustion conditions rather than on S-content or ash-content of the fuel.
Line 158-160: Isn’t there LNG as shipping fuel included in the technical ship descriptions (STSD)?
Line 164: Is Figure S3 based on IMO data? Please give the correct reference in the supplements.
Line -167-170: This threshold is very artificial and most likely not realistic. It seems unrealistic that only big ships use scrubbers while smaller ones don't. Please compare to existing scrubber installation data. Also, Figure S4 shows that HFO consumption is reduced with the size of the ships. Please also explain what you mean with MGO_5E3 and MGO_1E3.
Line 171-174: While the relative values match well in 2020, this gets worse in the following years. What is the reason for the increase in HFO in the IMO data after 2020? Is this caused by an increase in scrubber installations?
Line 179-189 and Fig. S5: It seems that for some ship types, no new ships are detected, e.g. RoRo, chemical tankers, … (Fig. S5). This may lead to uncertainties.
Line 196: “SEIMv2.3 primarily supplemented activities of ships constructed between 2000 and 2010”: So these are not “new” ships (i.e. recently built) as you stated before.
Line 204: Again, you talk about “new ship deployment”, while the age of most ships was between 14 and 24 years in 2024 as you state in line 196.
Figure 4: Why don’t you use simple bar plots? They would be easier to read. And why is there no increase in “Fleet number_IMO” by weight bin for ships other than < 10kt?
Figure 5: You should make clear that these are signals from ships that were not in your database before. It does not mean that these ships are new and it also does not mean that other ship emissions models did not include these ships.
Line 215: Which are these different models? Please introduce them here. EDGAR and CEDS are global emission models, not necessarily ship emission models. You should find out, which ship emission model is behind the data (e.g. STEAM).
Figure 6: Please explain STSD and FLSCCM in the caption.
Lines 222/223: Why is there no PM2.5 data from CEDS?
Line 232: In order to improve SEIM, it is not necessary that ships are identified on a daily basis. It is only necessary that they are not neglected and that the technical information about the is reliable.
Line 254/255: Indeed, there are multiple choices for being compliant with the global low sulfur rules. Your model, that assumes that only the biggest ships use scrubbers while smaller ones use LSFO or VLSFO may not be very accurate in terms of spatial distribution of PM2.5 and BC emissions.
Line 256-265: See my comment regarding line 232: I cannot follow the argument that a daily update is necessary when you run the model in a hindcast mode as described here (i.e. calculating the 2024 emissions some time in 2025 or later). Therefore, I do not see that Fig 7 and Fig. 8 contain useful information.
Line 275/276: Again, you should make clear that this increase of ships in the STSD does not mean that all of these are newly built ships.
Line 281/282: “by previous bottom-up method”: What do you mean here? Do you refer to SEIMv2.2?
Line 290: This “systemic emission underestimation” was only demonstrated for SEIMv2.2, not for any other ship emission model. You need to point this out.
Line 294/295: Why do you explicitly mention satellite studies for verification? There may be other possibilities.
Lines 305-307: You could test the implications of your assumptions for the spatial and temporal emission distribution. For example, you could check the size distribution of all known scrubber ships and assume that smaller ships may also use scrubbers.
Language:
Considerable improvement of the language is necessary, in particular, the use of articles and prepositions needs to be checked carefully.
Minor comments:
Line 10: what is a “hard-to-abate sector”? Please be more specific.
Line 14/15: What is “near-real-time” here? One hour, one day, ...?
Line 16: of the global ship fleet
Line 20/21: explain the “single compliance choice assumption”. This is a very model-specific term.
Line 83: explain STSD
Line 93: So you mean 113 million dynamic and 14 million static signals per day? Please specify this. Does this refer to a global coverage?
Line 174: in the Summary and Discussion section
Line 183: Does the percentage relate to the number or the DWT?
Line 183: Similar to before: Does the percentage relate to the number or the installed power?
Line 282: please correct “Most underestimation”
Check the References for errors, e.g. line 416
Citation: https://doi.org/10.5194/egusphere-2026-1935-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 158 | 56 | 16 | 230 | 21 | 11 | 12 |
- HTML: 158
- PDF: 56
- XML: 16
- Total: 230
- Supplement: 21
- BibTeX: 11
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This paper present an updated version of a ship activity model to better estimate the atmospheric emissions of CO2, SOX, BC, NOX and PM2.5 where both number of vessels, distance sailed and compliance option with respect to the IMO global sulfur cap is included. While presenting important new data, showing the underestimation of all emissions of atmospheric pollutants I lack a further discussion on 1) if authors think it is the adoption to daily resolution, the data curation and/or the addition of more vessels to the model that have the greatest influence on the results and 2) given the systematic underestimation (and 20% is a lot), what are the implications of the shipping fleet's impact on the envrionment and human health and what can be proposed as measures to improve reporting, data sharing and inclusion og global shipping in future climate/impact models? The authors focus on two compliance options (low sulfur fuels and HFO+scrubber) but given the high temporal resolution of ship activity dat, would it be possible to divide low sulfur options into VLSFO (in 0.5% areas) vs MGO (and ULSFO) (in 0.1% areas)? My guess is that emission factors are different but the question is if data is available to differentiate. On emission factors, would it be possible to present these as ranges instead of absolute numbers to reflect the variability? Is engine load accounted for in the modelling, if not the authors could discuss the implications of including this? For many pollutants, emisison factors are tested at different loads so it should be possible to do some estimations (present ranges as proposed in previous point).
Some minor comments: