This paper presents the results of an urban ecohydrological modelling exercise using the ECH20-iso model. Growing season field measurements, including soil water content were used to calibrate the model. Qualitative validation was carried out by comparing measured and simulated water isotopes, surface temperature, and simulated transpiration to measured sap flow data. Overall, the paper was well organized and enjoyable to read. The clarity and simplicity of the figures was appreciated. I think there is room to improve the clarity of the methods with some additional details (see suggestions in specific comments below). As well, I recommend the authors try to make the model validation more quantitative (see suggestions in specific comments below). Lastly, I think the authors could place some additional focus on the urban design/management implications of the grass results, which suggest they are as important as trees for regulating green water fluxes.

48-9: Reverse order of this sentence, i.e., … by increasing infiltration and groundwater recharge, and thereby reducing stormwater runoff.

55: Replace ‘removing’ with ‘to remove’.

57: Replace ‘gives recreational benefits’ with ‘provides recreational opportunities’.

57-8: Remove ‘…to actually…’

63: Should be ‘… quantities of water that are partitioned…’

70-1: The point being made here is unclear. I recommend revising the wording for greater clarity.

76: ‘setting’ should be plural.

81: Grammatical issue in this sentence – isotopes is plural, but sentence doesn’t reflect that.

96-7: Could the aggregated percentage of green and blue space be broken down for both?

Figure 1: Add a scale bar and north arrow to the zoomed in map of the observatory site.

144-5: Do the authors think that the drought conditions preceding the study could have impacted the results in any way?

151: Missing words in this sentence - ‘… installed at the top of the flux towers…’

150-7: What is the fetch (or footprint) of each tower? What proportion of the study period is each tower measuring convective fluxes that are representative of the observatory site? This will depend on wind direction of course.

Section 2.2.: Throughout this section it would be helpful to know the make/model of all the sensors that were used.

161-4: How was soil sampled? How was water extracted? How was isotope analysis performed?

166: I recommend providing a brief overview of methods and then referencing Kuhlmann et al 2020b for more detail.

188-9: Is available moisture at the surface not used as well for partitioning available energy into convective fluxes?

198: How is the vegetation rooting parameter obtained? Is it based on field measurements of rooting depth?

204: The calibration period is the period over which data were obtained from the site. Validation is qualitative. Is there 2020 data that can be used to validate?

206: The soil division in the model is unclear to me. Were soil surveys used to characterize the spatial distribution of soil types? If not, was there any verification of the assumed soil characteristics?

Figure 3b: Is the local drainage network storm sewers? Is there any channelized flow through the site?

239: I think you could use ‘ground-to-atmosphere’ instead of ‘upwelling’.

247-9: How common are “infiltration hotspots” at sharp interfaces between impermeable and permeable surfaces? In my experience, these areas are usually quite compacted, thereby preventing infiltration.

267-8: Were surface cracks visible to the field team?

274-5: Are the authors referring to the measured or modelled SMC here? Layer 3 is not shown in Figure 2. It’s not clear which data is being discussed.

299-300: I think the authors should try to provide some interpretation of the poor fit in late June, late Aug and early Sep.

307-8: What do the authors mean by ‘This compares with the 352 mm of precipitation during the calibration period.’? All the ET values are larger than P.

318-20: Is there some way to estimate E from these surfaces to see if they account for the missing amount?

324-7: In layer 1, SWC is generally higher at the beginning and end of the study period with fluctuations (but lower baseline) in between. Why is the water age relatively high in April/May and Nov? Maybe the April soil water is old from the previous winter, but November receives a fair amount of precip. Some explanation of the distribution of ages at the beginning and end of the study period would be helpful.

362-5: Would the authors recommend more, denser SWC measurements in urban soils?

391: Spelling error – trees.

396: It’s unclear what ‘… generally capturing processes adequately…’ means. What were the criteria for evaluating this? On line 397, ‘generally good reproduction’ is used. Can the authors be more quantitative, e.g., measured and modelled isotope values were within x-x % of one another? It’s unclear what is meant by ‘… this simulation required…’.

399: Similar to the last comment, ‘… somewhat less successful…’ seems too vague. Can the fit be assessed quantitatively?

399-400: I recommend being more explicit about the number of wetting fronts that the isotopic results reflected.

Figure 10: This is a nice conceptual figure.

Section 4.4: It might be worth mentioning that sometimes preferential flow pathways can form along impermeable-permeable surface boundaries. This would move water away from the area and potentially make it unavailable for green or blue water fluxes. Something to explore in future work perhaps. [I see the authors make this point later on line 454-6… excellent].

Section 4.5: Could the authors comment further on the role of grass in promoting green water fluxes? Many sustainability-focused landscape designers seem to be moving away from grass (or lawns), but perhaps the findings of this study are an argument in their favour.

The manuscript “Quantifying the effects of urban green space on water partitioning and ages using an isotope-based ecohydrological model” written by Gillefalk et al. provides a set of insights for water partitioning in a complex urban landscape. They incorporated the use of water stable isotopes in precipitation and soil to verify the model capacity for partitioning water fluxes. Also, they use eddy flux and sap flow data to evaluate the model results. Despite their meticulous work, there are some concerns about data collection and applicability.

Major Comments

• Authors mentioned in lines 147-150 the use of another urban flux tower for a portion of the sampling period. It is important to highlight the fact that despite their similarities as “Urban Environments”, the proportion of green spaces/buildings can affect considerably the model outputs. Also, the authors did not show a consistency analysis or a comparison for the period June – November in which both towers could be operating. This is fundamental to considerer the fluxes as similar, equal, or different. Fluxes such as outgoing longwave and shortwave radiation, as well as water vapor can be affected and give different proportions.  As an example, the 2-fold overestimation showed in Figure 7 with respect to sap flow data for April and May can be linked to the differences between flux towers without counting on the constant LAI effect.

              The authors should ensure that this data set can be used by this other location.

              * How different are the land covers within the tower’s footprints?

              * Are the fluxes for the period June-November equal/proportional/different?

              * Do both towers have the same setup in terms of instrumentation?

• The manuscript is based on the application of a model which requires a calibration period. The authors mention the application of this procedure (Section 2.4). Despite the detailed description of the calibration procedure, two main questions remain unanswered:

              * Which data (period and source) was used for the calibration?

              * Did the authors apply a spin-up procedure (how long) or not?

              This issue is important to assess possible trends or initial effects in the flux initial values.

• During the modeling procedure, the authors used a constant Leaf Area Index per cover. This can be true for the grassland depending on the species but the effects in trees and shrubs fluxes can be important. The application of this assumption triggered important consequences for the model results which end up with the overestimation of transpiration fluxes during the first part of the year (Figure 7 – April and May). However, the lack of sap flow data in shrubs affects the reliability of the fluxes from this cover.
• Along with the discussion, the authors mention the term “urban karst” given by Bonneau et al. (2017) which affects the water fluxes and redistribution by the preferential flow. Taking into account the heterogeneity of the subsurface on the sampling area (Lines 134-135) where the “subsurface is heavily impacted by human activities, and in places has an added layer of up to 50–180 cm of debris”, how does the preferential path flow form by these debris affects or potentially affects the soil water age estimations?
• How does the “tree with run on water” compare against the transpiration of the tree(s) sap flow measured in those pixels? This will support the affirmation given by the authors about the model performance and side effects of nearby impermeable land covers (e.g, buildings, pathways).

Minor Comments

• Does the soil water content measurements were calibrated with soil samples along the sampling period?
• The authors mention the use of a German Weather Station that “records essentially the same rainfall” (Line 156). Can the authors provide the values?
• What are the urban tree species sampled for this manuscript (Lines 159-161)? Can the authors provide more information about the individual trees sampled (e.g, diameter, species, height, etc)?
• The paragraph between lines 167 to 175 describes the results obtained from the data collection described in the previous methodological sections. Consequently, this should be in Results and not in Methods and Material.
• The authors mention the use of Nash-Sutcliffe efficiency (NSE) as objective functions (Line 216). However, across the manuscript, there is only one reference to NSE in a broader context (Line 234) with no reference to the results of this analysis and neither in the supplemental material. The authors only mention Kling-Gupta Efficiency in detail (e.g, Line 272, 282). What happened with the NSE analysis?
• - The authors should follow the recommendations given by Knoben et al. (2019) when using Kling-Gupta Efficiency analysis in models.
• - The authors should add more information about the results using the NSE. Also, adding the respective equations as for KGE.
• - It is necessary to add more information about the sampling processing (precipitation and soil samples). The following questions must be answered:

            * How were soil samples collected?

            * When collected, how long after a rain event the soil samples were taken?

            * Which soil water extraction procedure was applied?

            * What method/equipment/laboratory performed the stable water isotope analysis?

Recommendations

• The authors can use boxplots in “Figure 10: Soil layers” instead of bars. In this way, the reader can have a better idea of the data distribution for each layer/cover.
• The authors can add the transpiration envelop in Figure 7. This will allow the readers to have a better notion of the temporal flux variability.

References

1. Bonneau, J., Fletcher, T. D., Costelloe, J. F. and Burns, M. J.: Stormwater infiltration and the ‘urban karst’ – A review, J.Hydrol., 552, 141–150, 10.1016/j.jhydrol.2017.06.043, 2017.
2. Knoben, W. J. M., Freer, J. E., and Woods, R. A.: Technical note: Inherent benchmark or not? Comparing Nash–Sutcliffe and Kling–Gupta efficiency scores, Hydrol. Earth Syst. Sci., 23, 4323–4331, https://doi.org/10.5194/hess-23-4323-2019, 2019.