Here is yet another paper, of a growing number of papers, which shows us breaking free the of the simple models of fire behaviour of the last 40 years.

This paper makes several contributions first and probably most important is that this is a controlled burn i.e., a management burn of low intensity and low rate of spread. Its mesoscale weather affect is rather limited compared to most of the published studies of large area fires with turbulent sensible heat fluxes that penetrate much higher into the atmosphere.

In general, I found the paper very detailed and complete in its discussion and interpretation of the model and data. It is an excellent example of how to organize and evaluate fire models with specific research question. I suspect we will hear more of the studies related to this fire. These approaches are the future so we must recruit managers and researchers who do not necessarily have any experience or limited experience with this now more mechanistic approach.

It would have been useful for some more information on the distribution and density of the groundcover, shrubs and trees(height). Also on the  amount of necrosis, mortality and litter and duff removed.

General comments:

The work presented in this manuscript implemented and compared three different fire sensible heat representations in a mesoscale weather model equipped with a canopy module. The work focused on low intensity prescribed fire scenarios which presented a timely study considering the need of such knowledge and tools for the fire risk management in the prescribed burn practices. The three fire sensible heat source representations implemented in the ARPS-CANOPY mode will provide a great tool for future fire studies from mesoscale down to microscale.

Few efforts have been made to cross-compare and assess the suitability/limitation of these methods. Another important contribution of this work is the evaluation of these different fire representations. Utilizing extensive observational data from a field campaign, this work is able to evaluate different fire sensible heat source representation through sensitivity tests and the model-observation agreement. 

In my opinion, this is a well written manuscript which presents not only the advancement in the numerical representation of the fire sensible heat source but also with sensitivity tests that provide guidelines for the future studies using the modelling tool. That being said, a few minor revisions might help the readers to better appreciate the work. See below for the minor issues.

1. Line 50 to 56.  I think the concept of the mesoscale model used here is slightly different from the traditional definition. Traditionally, most of the mesoscale model applications were done using pure parameterization with little or no resolved turbulence like Reynolds-averaged Navier–Stokes equations (RANS) rather than large eddy simulation (LES) approach. Recommend the author to add one or two sentences to clarify.
2. Table 1. The vertical extent/size of each domain should be provided. 
3. Figure 5. The use of “x” in the figure to indicate minimum and maximum caused overlapping of the symbols in some of the subplots. Recommend to use a smaller “x”. A simple figure legend to highlight the observation versus modeled data might also be useful.