Dear authors, this is a carefully prepared manuscript and there is much need to enhance PALM capabilities with respect to particle emissions and transport modeling, given the importance of air quality modeling in urban physics models. I have some fundamental questions that can be addressed to result in a better manuscript. These mainly relate to key assumptions about the source of particle emission and the physics of their dispersion.

Page 3: Equation 1:

PM10 particles contain large particles with diameter > 5 micrometers. Such large particles can experience gravitational settling, which requires another term in the transport equation (1). I would say the the transport equation applies to smaller particles within PM2.5 that do not experience gravitational settling. Is it fair to say that the modeling approach is more suitable for PM2.5 than PM10? If you agree, then the model should be advertised as a transport model for PM2.5. If larger particles are to be accounted for correctly, then a gravitational settling term should be added to the transport equation.

Section 3:

Particles emissions from domestic heating are only present when fossil fuels (e.g. natural gas or diesel) are used which result from combustion processes in HVAC furnaces and water heaters. However, many buildings rely on heat pumps for heating, which only consume electricity. In such a case there will not be particle emissions from heating. Does PALM allow for such heating technologies? If so, particle emissions from heat pumps should be set to zero.

Equation 12:

This equation is used to consider heating requirement for buildings as a function of Toutdoor-Tsetpoint, however, I think it is more appropriate to use heating degree days instead. Even in the absence of a building energy system, Tindoor can be higher than Toutdoor due to internal heat gains (people, equipment), conductive-convective-radiative heat transfer, and thermal inertial of buildings. An energy balance in the absence of building energy systems results in degree days. Would it be more appropriate to estimate the building heating needs using the degree-day approach, rather than Toutdoor – Tsetpoint one can use Toutdoor-Tindoor in the absence of building energy systems? In summary, fundamentally, degree-days (Toutdoor-Tindoor) estimates are different from temperature deficits (Toutdoor-Tsetpoint). Weather and climate models calculate degree-days considering an energy balance model between buildings and the outside environment in the absence of a building energy (heating/cooling) system.  

General Comment

The paper provides a technical description of the new emission module for PALM that allocates discrete emission sources as volume source terms in the prognostic equations. The functionality of the emission module and the emission parametrization based on the temperature deficit between the ambient and user-defined target temperature is tested for PM10 emissions from domestic heating sources for a heterogeneous urban area containing many building stacks. The spatial association between the individual emission source locations and the corresponding cell index locations in the computational domain are maintained using a hash map for each emission sector, which offers an elegant way for the book-keeping of emissions. Vertical profiles of PM10 concentrations due to domestic heating, resulting form a 48-hour test simulation, were not validated against measurements. This deficit of the study may be excused by the focus on presenting the emission module under idealized conditions. Nevertheless, it is strongly recommended to perform tests treating domestic emissions as either buoyant or nonbuoyant volume sources. The heating exhaust should be a warm plume, which rises by buoyancy, especially in winter when air temperatures are low. The buoyancy effects might be less effective than the turbulent mixing, however, this needs to be investigated for the conditions of this study. It is referred to a study by Langner and Klemm (2011), who demonstrated that dispersion models work acceptably for nonbuoyant volume sources, but don’t cope with buoyant volume sources. Another aspect of PM10 emissions from domestic heating is that they are partly volatile. Residential emissions of organic carbon are largely semi-volatile and intermediate volatility compounds (S/IVOCs). The authors should explain how the modular emission concept can be extended in the future to handle the volatile fraction of emissions and incorporate the emissions of S/IVOCs. They should also discuss the representativeness of the meteorological conditions in the 48-hour simulations for the winter period.

Specific Comments:

1.) Introduction (P3, line 58-61): The two examples (trees and exhaust emissions from aviation) given here do not have much in common. Which vertical resolution is meant in relation to trees and aircraft? Approaching and starting airplanes emit in a height up to 900 meters within several kilometers around airports, potentially affecting ground concentrations. Further, the phrase “sufficiently low horizontal resolutions” sounds strange, as models generally try to achieve high resolution.

2.) Introduction (P3, line 64-65): Volume sources are a quite common way to treat diffusive sources in dispersion models. Mention how other models for the urban scale, e.g. AEROMOD and AUSTAL deal with (nonbuoyant) volume sources.

3.) Model description (P5, line 123-125): While the hash map is described as a clear connection between the emission database and the cell coordinates (i,j,k) where the emission of a source is added to the prognostic equations, it is not clear what happens for different cell sizes and volumes of the defined grid. How is it assured that the emission source is allocated to the correct cell when the grid cell size and volume is changed in the model configuration?

4.) What is the footprint of a building and how is it calculated (P8, line213)?

5.) Module implementation: It is not clear how the height level of the building stack is considered. The module implementation section only mentions the (i,j) cell location of each building stack. The volume source is probably defined at the height of the stack exit and not the entire building is the volume source. Are there any plausibility checks of the user-provided emissions? There should be some internal control in the emission modules that check the plausibility of finally calculated emission rates and gives warnings when emission rates are unrealistic or not defined.

6.) It would be interesting to see a more generalized approximation of the vertical profiles shown in Figure 9 for sampling sites A-F as time average, for example in steps of 10 m above ground. The average vertical profiles should be compared to more generic vertical profiles of heating emissions in urban areas found in the literature.

7.) Define the reference height (P16, line 476). What causes the vertical mixing of heating emissions, does buoyancy of the heating plume play a role here or not? The occurrence of down-wash and accumulation should be explained in terms of meteorological conditions, not only in terms of trapping in building enclosures.

8.) Figure 10: in top row (A) the vertical cross-section shows a hotspot at 20:00 at around 30 m, despite there seem to be no emission stacks of buildings close-by.

9.) The Concluding remarks should address the limitations of the domestic emission parametrization. The uncertainties of the emission factors are large and cannot be ignored. Also the diurnal variation in domestic heat usage can be locally different from the one defined in CAMS for other stationary combustion.

Reference:

Langner, C., and Klemm, O.: A Comparison of Model Performance between AERMOD and AUSTAL2000, J. Air & Waste Manage. Assoc., 61(6), 640–646, DOI: 10.3155/1047-3289.61.6.640, 2011.

Review of "An enhanced emissions module for the PALM model system 23.10 with application on PM10 emissions from urban domestic heating"

by Edward C. Chan et al.

The article describes the implementation of an emission module in the PALM model. A performant and flexible emission model is an important contribution, enabling a broad range of studies with the model, including air quality modelling. However, I find that the description of the implementation needs improving.

In the review, I focus on the implementation description in section 2. 

In general, what is meant by "hash map"? The relation in Eq (3) and (4)-(6), or an array (or other data structure) of emission sources indexed by kappa? 

What I miss is a description of the data structure of emission sources. How is it organized? For a given cell (i,j,k), how are the emission sources in this cell found? (i,j,k) maps to a single kappa value. How are the emission sources found then?

If the array of emission sources is shorter than the number of grid cells, there should be a search operation to find the sources corresponding to a given grid cell. If the array of sources has the same length as the number of grid cells (which makes look-up easy), it seems one could just use a full 3D field of source strengths instead, with the same memory cost but less complicated code.

A comment of how the implementation handles domain decomposition would be helpful. The list of emission sources is presumably prepared for each MPI process?

line 90 (language):

"associated *from* this approach"

line 91 (language):

"must be so conceived accommodate them"

line 97:

"3. The interface between the prognostic equation solver and the emission module should be implemented to allow only localized data access to prevent propagation of data corruption into other emission sectors."

I don't understand this statement. Data corruption would be an error in the implementation. Isn't it an obvious design objective that the implementation should be error-free?

Eq (2):

the notation feels unnecessarily convoluted, with the W sets with multiple indices.

Additionally, in "W ∈ 0, 1, 2, · · · up to the corresponding upper bound N"

presumably the set does not include N, but this is not clear from the formulation.

The mapping in eq. (3) is quite trivial, just enumerating all the grid cells. Usually a hash function implies something more, e.g. that the output space is smaller than the input space (although this is no strict of formal requirement).

Eq. (4) is wrong, it should have a division not mod. Additionally, there is an implied rounding down after the divisions,

which could be indicated with a floor function or with round-down vertical-bars-with-hooks symbols

I don't understand Eq. (8) or the explanation above it. Additionally something is missing in the sentence "...p is the union all emission sectors".

Eq. (9) What's meant by the union of hash maps?

130: "Each *mdoule*"

200: "functions and subroutines that can *bed* used in other emission sectors."

line 268: "specifies the annual cumulative temperature (in degrees) to be heated above the ambient temperature to the target temperature, with a default value of 2100 K."

It's not obvious what annual cumulative temperature is. If it is something

like degree days, the unit is wrong.

498: "an test case"