Find the review in the attached .pdf file.

In this study the authors focus on the NW South America region to investigate the 3D distribution of the crustal seismogenic thickness as well as the temperature of occurrence of crustal earthquakes. They propose an integrated workflow that includes a lithological and a thermal model and incorporates observed seismicity. This approach is particularly suitable for this complex region, where different tectonic plates interact, there are several subduction zones, thick sedimentary basins and accreted terranes. An important contribution of this kind of study is that it improves the understanding of seismogenesis and has implication for future seismic hazard assessment. This is particularly relevant in the study region, which is seismically very active and has hosted devasting earthquakes.

The manuscript is well written and organized, but my main problem is that I found that many statements were poorly discussed or justified. Similarly, I often could not see in the quoted figures the statements made in the text. Some important conclusions were not sufficiently supported in the text. For example, the authors find that the hottest domains correspond to the deepest values of D90, without giving any physical explanation, and from this correlation they make the conclusion that ‘The spatial variation in the geothermal gradient in the uppermost 20 km of the lithosphere has significant predictive power for forecasting the distribution of seismicity’. First, I find that this correlation between high geothermal gradient and deep crustal seismicity deserves further validation and discussion, since it is, at least, counterintuitive. A warmer crust should promote a ductile instead of a brittle (and then seismogenic) behaviour, for this reason the brittle-ductile transition should move to shallower depths and then the base of the seismogenic crustal layer should be shallower, what is the opposite to what obtained in the present study. I consider that this issue and other comments that I elaborate in the following notes can be addressed in a moderate revision.

Main comments:

The last two paragraphs of the Introduction section have a level of detail about the adopted approach that the reader is not ready to follow at this point of reading (for example the validity of the steady-state assumption…) I suggest keeping this part more general and express the main purpose of this study. The style of the last statement (‘the main advantage…’) is more appropriate in this regard.

Lines 77-78. ‘In order for this approach to be realistic, our thermal model considers only the uppermost 75 km of the lithosphere’. The authors should justify this choice and explain in what sense it is more realistic. The depth of 75 km is shallower than typical lithosphere-asthenosphere boundary (LAB) depths.

There is not any discussion on the possible effects of the steady-state assumption. The authors argue (lines 75-77) that a steady-state approach is appropriate since they target crustal earthquakes. However, transient effects can also affect the shallower parts of the lithosphere. The authors also mention that the subducting segments of the Nazca and Caribbean slabs are flat in the study area. This is not a direct support for the steady-state assumption and in addition is contradictory with the steep Coiba slab plotted in Figure 10.

Line 175: ‘All lateral borders of the model are assumed to be closed’ I don’t understand this condition. Do the authors mean that the borders are insulating, with a zero horizontal heat flow?

Concerning the structure, section 3.1.3 ‘Validation of the modelled temperatures’ mentions the method used for this validation, but I do not see how this validation is actually performed. Related to this, I don’t see the purpose of section 3.1.4 Geothermal gradient. Can the authors briefly advance the purpose of this calculation?

Lines 190-192: the statement :’ the geothermal gradient for the crustal seismogenic zone was computed as the temperature difference between the surface and 20 km depth’ seems to me just contradictory with what said just few lines before about the geothermal gradient being computed every 3 km. It seems therefore quite inaccurate just using the temperatures at the surface and 20 km depth.

Line 194. The statement ‘A similar approach was followed by Gholamrezaie et al. (2018)’ seems rather ambiguous? Do the authors refer to the approach to compute the thermal gradient of the 3D thermal modelling? Please clarify.

Line 285. The resulting D10 and D90 values, and their corresponding standard deviations are provided in the data repository (Gomez-Garcia et al., 2022). Why not shown here in a map?

Line 299. The authors mention that ‘it is possible to conclude that the model (heat flow) fits the regional trend’, but honestly, I don’t see this satisfactory fitting in figure 4b. Similarly, modelled temperatures shown in figure 4a seem to systematically underestimate observations at depths < 4 km. Can the authors hypothesize about this (small but systematic) misfit? 

Lines 320-321. I don’t follow the logic of the argument ‘This again is an indication that a 1D approximation is not robust enough to model the thermal configuration of the study area’ as the occurrence of earthquakes in regions with a diverse range of geothermal gradients would be also reproduced in 1D or 2D models, although these 1D or 2D approaches become less accurate in areas with large lateral thermal variations.

Lines 32-327. I don’t see that in what sense the correlation between seismicity clusters and high geothermal gradient suggests that this geothermal gradient occurs in ‘places where the crust and lithospheric mantle are strongly coupled’. Theoretically, a high temperature should lead to the shallowing of the brittle-ductile transition and therefore it should promote decoupling between the upper crust and the lithospheric mantle, with a ductile-aseismic lower crust in between. In summary, keeping the rheological stratification of the continental crust in mind, a strong coupling between the crust and the lithospheric mantle should correspond to a colder crust, and therefore a predominant brittle behaviour and absence of a ductile layer in the lower crust weak enough to mechanically decouple crust and lithospheric mantle.

Lines 458-459 ‘Instead, the hottest domains are associated to sedimentary basins (Fig. S3) and correspond to the deepest values of D90’. I think that a simple interpretation for this is that temperature increases with depth and therefore deepest values of D90 occur at higher temperatures. This seems in contradiction with what said in lines 510-513 ‘However, we observe a general trend between the lithospheric configuration and the seismicity distribution, that is the colder and therefore stronger the lithosphere, the deeper and higher in magnitudes the earthquakes (e.g.: Chen et al.,2013).’ Overall I do not understand why a cold lithosphere correlates with deep seismicity, while a high geothermal gradient correlates with deepest D90. I do not see a plausible explanation for this opposite effect of the thermal state in the crust and in the lithospheric mantle on seismicity distribution.

Lines 475-476 ‘a thick lower crust together with a relatively hot upper mantle could contribute to large hypocentral temperatures (Sect. 4.5)’ I do not understand the relation between the lower crust thickness and the hypocentral temperature. As in my previous comments I suggest the authors to give an explanation for this kind of poorly explained inferences.

lines 510-513 ‘However, we observe a general trend between the lithospheric configuration and the seismicity distribution, that is the colder and therefore stronger the lithosphere, the deeper and higher in magnitudes the earthquakes (e.g.: Chen et al.,2013).’ I don’t see this in Figure 10. Please, note that deep earthquakes beneath the Coiba slab correlate with a shallowing of the 600 ºC isotherm.

Line 579 ‘the seismogenic crust is thicker and hotter below the thick Middle Magdalena basin’. I see in figure 10 that the seismogenic crust is thicker, but not hotter than beneath the surrounding Murindó and Guaicaramo/Yopal faults.

Minor comments:

Line94: remove the typo Reguzzoni & Sampietro,ta 2015

In color code for figure 2a, better say ‘lithospheric mantle’ instead of ‘mantle’. Similarly, in the second line of the caption of Fig. 2, better say ‘lithospheric mantle’ instead of ‘upper mantle’

Line 392. The reference to Fig 2 is wrong

Line 551 ‘for calculating of the thermal field’ remove ‘of’