General Comments:

The authors present an enhancement to the JULES fire model INFERNO, which enables that model to simulate peat fire (ground fire) in northern latitudes.  INFERNO-peat is implemented in Python as an external model that is driven by output from JULES-INFERNO.  Simulations driven by two different ignitions forcing datasets for 1997 - 2014 were compared to several global and one regional observational datasets.  The INFERNO-peat implementation increases northern latitude burnt area estimates bringing it closer to observation products such as GFED5 and FireCCILT11.  The spatial correlation with observations is reasonable, but with some biases.  The simulations also had larger and more realistic intern annual variability than INFERNO.  Emissions were correspondingly increased with INFERNO-peat, closer to total emissions estimates.  Again regional biases were present.  The GFED 500 meter product, which contains aboveground and below-ground emission estimates was used to look at emissions more closely.  This revealed that biases were dominated by below-ground carbon emissions being too low or too high in different regions.  Inaccuracies in estimates of vegetation in JULES-INFERNO are likely propagating through to drive some of the biases in INFERNO-peat as well.  Improving prediction of peat fires in models is important because of the both climate and societal impacts.  Improvements to INFERNO-peat could come from better representing human behavior and the specifics of peat emissions.  The authors also highlight the need for better detection and estimation of ground fires, which is not always observable with current satellites.

The authors have presented a well written and interesting article documenting INFERNO-peat.  The model extension is welcome advancement and the simulations results are compelling.  This work is relevant and novel.  I can only suggest a few minor changes and areas for improvement as follows:

In Figure 1. the ‘r’ is missing in ‘moisture’.

The equations in section 2 would be easier to understand if units were given with the first mention of each variable.  While the units for some are implied or inferable, others are not.  Specifically, I could not determine the units for FlamPFT in section 2.1, soil moisture (SM) in 2.2, and burnt area (BApeat) in section 2.3.

In figure 3 the color scale makes it very hard to make out any signal in Fennoscandia and Alaska for most panels.  Changing the scale so that true zero cells have their own color (white?) while low non-zero values have another color would make comparisons easier.

Figure 9 is interpretable with aid of the text.  However, it would be clearer if the INFERNO bar was marked as “aboveground”  and the peat only as “below-ground” in the legend.

The discussion is quite complete.  However, a few lines discussing any potential benefits of (or obstacles to) full integration of the INFERNO-peat logic into JULES-INFERNO would be appreciated.

Given the number of colors in supplemental figure 5 it would be helpful to either mark any PFTs not shown in the plot or remove them entirely from the legend.

This work developed a new model INFERNO-peat. It has some improvements from the original INFERNO model in terms of estimations of burnt area, carbon emissions, etc, especially in northern high latitudes. The major comments are summarized as follows:

1. How did the model consider the effects of wind speed and ambient temperatures? I believe they play important roles in the ignition and spread of peat fires.
2. L55, ‘‘but can burn to as deep as 50cm’’. Some recent lab experiments showed it can burn at 100 cm depth (Qin et al. 2022)
3. L150, from Eq. 2, the combustibility of peat soil only depends on its moisture content. Even though the authors state MC plays dominate role, there are many studies emphasizing the significance of other factors like inorganic content (Frandsen 1997), ambient temperature, fuel density, etc.
4. Line 165, the unit is missing.
5. Line 180, the carbon emission calculation is too rough. I understand the authors try to calculate the total emitted carbon. But your cited works either use emission factors or carbon emission flux (7.1 kg C/m²). Assuming that all carbon (C) from the burned fuel (Eq. 5) is completely converted to emissions is far from realistic.
6. In Fig. 2a, it indicates peat becomes incombustible when MC =100%. But in Line 48, it states “However, fires can still be maintained at moisture contents as high as 160% (Rein, 2013; Hu et al., 2019b; Rein, 2015; Purnomo et al., 2020)”. It is because the critical MC can change with other parameter (Frandsen 1997)
7. 3, It seems INFERNO-peat can capture more fires in high latitudes but less fires in low latitudes (compared with GFED and fireCCILT11), especially in Eurasia area. Can the authors explain why?
8. 5, 8, 9: The authors compare the average values over several years. However, providing subplots with the average values for each region on a yearly basis would be more compelling.

Frandsen WH (1997) Ignition probability of organic soils. Canadian Journal of Forest Research 27, 1471–1477. doi:10.1139/x97-106.

Qin Y, Musa DNS, Lin S, Huang X (2022) Deep peat fire persistently smouldering for weeks: a laboratory demonstration. International Journal of Wildland Fire 32, 86–98. doi:10.1071/wf22143.