General comments

This paper demonstrates the effect that modelling groundwater in the CABLE land-surface model has on droughts and heatwaves, using two droughts and multiple heatwaves in South East Australia as case studies. This is a very important and topical issue, and well within the ESD remit. The analysis is thorough and well-designed, and described in sufficient detail to allow reproduction. Particular attention is paid to understanding the mechanisms behind the results, which considerably strengthens the conclusions. The relevance to climate model projections is particularly well put, clearly stating the important implications of this study whilst carefully outlining uncertainties and avoiding over-generalisation.

All of the plots in both the main manuscript and the supplementary material are important for the arguments presented, and of a high production standard. The prose is well written, the structure is good, and there is a high attention to detail.

Overall, this is a very strong manuscript, which will make an important contribution to the field.

Specific comments

Section 3.1: This section shows that CABLE-GW has a very good agreement with GRACE total water storage, and that the GW run has a better agreement with GRACE than the FD run. However, I’m not completely convinced by the conclusion that the underestimation of TWSA in the FD run is because of the lack of groundwater representation. (i.e. I don’t think that other model deficiencies with the potential to reduce E have been ruled out). This is particularly the case as the accumulated P-E in GW run is still substantially different to GLEAM, showing that there is still an issue with the model even with groundwater included. To address this I would suggest (a) being a bit more cautious in the phrasing  so that the text doesn’t imply that including ground water is the only way to make the model match more closely to the observations and (b) including some text discussing possible reasons why the GW and GLEAM lines do not agree in figure 1b.

Line 229: “FD underestimates the magnitude of monthly TWSA variance (standard deviation, SD = 37.18 mm) compared to GRACE (47.74 229 mm) or GW (47.67 mm)”: consider showing this explicitly in a plot, as it’s an important result, which is difficult to read off Figure 1a.

Line 351: elaborate on why MODIS LST is lower than all the model lines in Figure 6h, even DR.

Line 369: “first ~two years of a multi-year drought”: link this explicitly to plots (as far as I could see, this is the first time this was mentioned, but it’s picked out as one of the main points of the study in both the abstract and the conclusion).

Line 398-9 “Our regional based results support this hypothesis and in particular highlight the importance of groundwater for explaining the amplitude of fluxes in wet regions (Figure 1)” Elaborate on how Figure 1 shows this.

Technical corrections

Line 434-436: This sentence doesn’t read well. Is it missing a “that” or an “and”?

In this study, Mu et al. evaluate the contribution of groundwater (GW) to vegetation water availability during heatwave and drought events in SE Australia. To do this, they implement a GW scheme in the CABLE land-surface model, and perform factorial simulations constrained by LAI to separate the contribution of GW to evapotranspiration and canopy temperature, which they compare with remote-sensing data.

The manuscript is concisely and clearly written, well structured and appropriately referenced and provides an important contribution to the land-surface modelling community. I find that two aspects could be improved:

(i) After reading the title one would expect a greater focus on the impacts of GW on vegetation functional aspects (assimilation, stomatal conductance, transpiration, growth…), while the manuscript focuses mostly on hydrometeorology. Transpiration differences between the two simulations are only shown in Fig. 2, for 2019, but they could have been included in Fig. 3 and S6, to complement the discussion about the functional aspects. From the model simulations, one could additionally include, assimilation rates, stomatal conductance, NPP, etc.  Moreover, the DR experiment is one of the most exciting aspects of the study since it highlights the relevance of the interactions between hydrology and physiology, but it is briefly discussed and shown only in Fig. 6. I find that a deeper analysis of the DR experiment and an additional figure on the impacts of the heatwaves for GW, FD and DR would increase the relevance of the study and better support the discussion around improvements to LSMs.

(ii) one of the key conclusions mentioned in the abstract and sections 4 and 5 is that GW helps sustaining higher transpiration rates in the first 1-2 years of multi-year droughts. However, most figures of the paper, and specifically the one showing transpiration differences, refer to the 2019 event. The figures showing differences between GW and FD for the full period do not separate specifically T from ET. Moreover, even though slightly bigger differences between GW and FD are seen in 2002 (1 year following drought onset) and 2017 (drought start), strong differences are found also in non-drought years, e.g. 2013 (Fig. 3 and S6). I do not think that the results, as currently shown, can support strong conclusions about the duration of the effect of GW on HW effects.

Other comments:

L100: I think dimensional analysis gives the units of Fsoil as m3.m-3/s, or 1/s (to match the other two terms), can you confirm?

L235: “much closer”: indeed, but still very far.

L250: can be complemented by a map of root length in CABLE.

L312-317: how much is this threshold dependent on model structure and parameterization? And how does it compare with the same results for the DR experiment?

L325-331: very hard to compare panels a-b with c-d. Can you use a consistent mask?

L339-341: the label of Fig. 6e,f indicates “GW-FD”. Can you check? I would find it important to show DR in more figures, as discussed above.

L343: how can one compare panels a,b with g,h in Figure 6?

L375-376: can you support this by separating results per WTD bins rather than simple visual inspection?

L376-377: Where can we see the time-dependence of this response? 

L408: specify what feedback is meant here

L448-449: rather than making a general statement, the authors could analyze variables related with the physiological responses (assimilation, stomatal conductance, WUE, NPP) to show that (if) GW matters.

L486: what can we see a figure supporting this conclusion?

L487: the cooling effect is shown only for 2019, correct?

L490: this is not strongly supported by results (comments above)

Please find enclosed my review as a pdf file.