General comment

The paper present a new modelling approach to take into account movements of ships in simulation of transport and dispersal of atmospheric pollutants. Some specific cases have been simulated and compared to other modelling approaches such as fixed emission points and line source models. The paper is interesting, suitable for the Journal and could be of interest also for future studies. However, there are several aspects that could be made more clear and also an over-interpretation of results because I believe that it is not very clear that this new modelling approach is performing effectively better compared to other approaches at least when several ships are considered (see my specific comments). In conclusion, I believe that the paper should have a carful revision before publication.

Specific comments

One aspect that raised my curiosity is why SO2 has not been considered in this work. Shipping-related SO2 is quite important and the most recent international policy from IMO enforce the sue of low-sulphur content fuel that will have a strong impact on the emission of SO2 (in addition to particulate matter). In addition, it is mentioned that simulations were performed also for PM2.5 but I only see NO2 in the results.

In several parts of the paper there is a confusion between “emissions” and concentrations”. In line 154, 162, 202-205 I believe that authors are mentioning actually concentrations rather than emissions.

Lines 171-175. These limitations are effectively necessary? What are the reasons behind these choices?

The simulations cover a very limited period (only a few hours). Will the results of the comparison among the different models similar if larger time span are used for simulations? Often daily or seasonal averages are used to investigate the impacts of specific sources to air quality. I ask this because, it seems that when a large number of ships are included, the differences of moving ship model and line emission model become negligible.

 What are the emission conditions such as vertical exit velocity, height of emissions, buoyancy and so on? Are the same conditions used for each ship or a difference has been done according to the typology of ship (for example cargo, cruise, ferry and so on). Could author comment if the uncertainty arising from the assumption made on emissions is smaller or comparable with the differences observed among the different models?

Figure 7. It is actually difficult to compare results because the colour scale is very different. It should have been better to have all figures with the same colour scale. The same for Figs. 13, 14, 15

Figures 16 and 17 are based on only a few points so that the correct line joining the points is a straight line rather than a “non-estimated” curve.

Lines 349-360. The better performance of MPS is not really visible. Results of MPS and line source in Figure 17 are essentially the same with negligible differences especially if compared to the uncertainty rising from assumption on emissions. Therefore this part of the discussion should be revised.

The same thing for the conclusions (line 370-373), it is possible that, for very short calculation periods and with a very limited number of ships, MPS could furnish more realistic results. However, this aspect is not really demonstrated by comparison with measurements. Regarding the conclusions (lines 383-387) of “real-world” simulation, I believe that it is not true that MPS furnish better results than line source model. The differences found among the two approach are negligible. This should be clearly stated in the conclusion of the paper.

The paper presents the development of a moving point source (MPS) model and compares it with two other common methods for evaluating the local dispersion of ship emissions, namely the line source model and the fixed point source model. The MPS was implemented as sub-grid module in the urban chemistry transport model EPISODE-CityChem to study the impact of ship emissions in numerical experiments with one and several ships, as well as in a real-world scenario in Singapore. The simulations are carefully done and the results properly discussed. MPS has a great potential when used together with AIS ship position data for real-time simulation of pollutant dispersion from ship emissions. The moving point source model is a valuable addition to the EPISODE model for assessing impacts of ship emissions on air concentrations and human health on local and city scale, allowing for more details on the spatial and temporal distribution. The comparison between MPS and the line source model reveals that differences between the two methods are more obvious for instantaneous concentrations than for longer averaging times (1 h), because the dispersion of single plumes released at different points along the same trajectory (line) becomes more homogeneous in space when longer time scales are considered. The real case did not reveal the clear benefits from using MPS, probably because the period of five hours was too short to cover different weather conditions, changes in the boundary layer structure or day/night variation. My suggestion is to include one more case that studies the role of different atmospheric stability conditions on the ship impact from the three emission models. Overall, I am in favor of publishing the paper after my specific comments below are sufficiently addressed.

Specific Comments:

1.) The abstract should better reflect the results from the study and give quantitative information about the evaluation of the performance of MPS. In particular, the larger discrepancies between the emission source models for instantaneous than for averaged concentrations should be stated more clearly.

2.) Is there any specific treatment when the plumes released from different virtual points during a simulation hour are crossing or overlapping each other? There should be some assumption about merging of the plume masses or other interaction between the individual plumes.

3.) Figure 4: what is causing the structures in the wind field?

4.) In general, there is too little information about the plume rise algorithm. How is plume rise handled in the LS model?

5.) What is assumed about the ship building height and width, since they can influence the plume rise?

6.) Real case study, section 3.3: a longer simulation period could reveal discrepancies between MPS and the LS model. The differences between the two models are very small and based on the real case it is currently not possible to conclude that MPS performs better. It is suggested to show the average 2-D fields for the observation period, to analyze where the largest discrepancy between the models occur and to look at a time series in the place of largest impact from ships. Differences in hourly average NO2 concentrations can be noticed, for example when looking at the 2-D plots in Fig. 15 at 180 min simulation time, over the eastern part of the city. It may be considered to show 2-D maps of concentration differences, to make such details clearer.

Technical Corrections

1. 10 line 20: emission and concentration are confused here.
2. 18 line 328: mention that locations of observation stations are shown in Fig. 6.
3. 18 line 330: replace “may contributed from following aspects” by “may be attributed to the following aspects”.
4. 19 line 339: replace “very” by “vary”.
5. Figure 15: MPS and LS model 2-D plots ought to have the same color scale to allow for a quantitative comparison. This also concerns the other 2-D plots, like Fig. 13 and Fig. 14.