Comment #3: I'm missing an aquifer water balance, this can be very useful to put the recharge and pumping values in perspective and to assess sustainability of the system under different scenarios.

Response #3: Thank you for another great suggestion. We agree that the addition of aquifer water balance would be very useful to visualize the propagation of these two variables (recharge and abstraction) to the storage projection. A new figure on this visualization with its description is available in Section 3.4 (Groundwater storage projection). We also discuss it in Section 4.2 (Impact assessment on future groundwater level projection), as we can see from the additional new Figure 9 (also attached as a supplement to this response), that these times (between the years 2020 and 2025) are the crucial time as volumetric-wise, the annual groundwater abstraction is estimated to be at the cross-section with the total annual recharge. Once the groundwater recharge volume as the main inflow has been breached, it is going to be way more difficult to restore the basin’s groundwater storage condition, despite other fluxes involved (surface–groundwater interaction, for example).

Comment #4: title: it's not the projections that will impact water availability, so better to change "future climate projections" to "future climate change" or "future groundwater recharge" (unless you actually mean that the projections will lead to decisions that will impact groundwater availability). Also, I would suggest to change "anthropogenic activities" to something more specific like "groundwater abstractions" or similar.

Response #4: Thank you for your suggestion. We agree to change the title of our manuscript so it can describe the content more precisely to: ‘The impact of future changes in climate variables and groundwater abstraction on basin-scale groundwater availability’.

Comment #5: line 101: to what extent is the aquitard spatially continuous?

Response #5: Thank you for your question. Despite not being actually 100% continuous within the whole groundwater basin boundary, such a setup (of continuous thin aquitard layer) shown in the conceptual model (Figure 1b) is the dominant setting indicated in most of the data that we have. We derive our interpretation of the aquifer layering based on borehole data distributed across the basin, collated in previous studies (Rahiem, 2020) – mentioned in Section 2.4.3. We reported in our previous studies that these data are not uniformly distributed, especially in highly elevated areas. To fill in the gap of knowledge on the aquifer layering profile, we derive the conceptual model parsimoniously, and verify the assumption with the simulation results. This way, not only do we tackle the data sparsity by comparing the simulation results both qualitatively and quantitatively (reported in our previous studies), but we also verify our model assumption to define the aquitard as a continuous layer within the calculation. Despite a minority of the borehole data (~ 5%) does not show the aquitard layer in its soil stratigraphy, we believe our approach to simplifying the model setup in this setting is optimum in representing the basin subsurface physical characteristics. We do not explain all of this in the manuscript, because it is pretty extensive (this will take at least 15 to 20 additional lines), and it has been mentioned and discussed in our previous papers that we refer to in the manuscript.

Comment #6: figure 1a: can you explain how the basin was delineated? is it based on topography?

Response #6: Thank you for your question. In answering this (and the upcoming) question, first we apologize if we refer the answer a lot to our previous studies. To provide a short answer to the question, the groundwater basin shown in Figure 1a is delineated by considering two sources: (1) the official Indonesia government document on the boundaries of the groundwater basin and (2) surface topography MERIT-DEM (Yamazaki, et al, 2017). In the first document, it is stated that the groundwater basin is delineated based on the subsurface data that are collated by the government. Further information on the raw data used to delineate the basin was, unfortunately, not available/accessible. To verify its accuracy, we compare the groundwater basin delineation with the surface catchment area, and they are very similar (again, the comparison of these two boundaries was shown in our previous paper). To provide some minor adjustments considering some context and logical alignment between the surface and the subsurface basin, we incorporate the surface topography from the MERIT-DEM products to fine-tune the final groundwater basin boundary that we use in this study. MERIT-DEM is chosen as the benchmark for the sake of consistency, as we also use MERIT-DEM in our hydrological simulation using Wflow_sbm model.

Comment #7: line 150: one-way coupling is justified if water tables are relatively far below land surface, is that the case here?

Response #7: Thank you for your question. In our test basin, the surface elevation variation is large, causing the water table to also have a large variation. The lowest point in the basin is found at around 600 meters above sea level, while the highest point at around 2400 meters. Therefore, in the mountainous area, the water table is relatively far below the surface up to several tens of meters. On the other hand, in the lower elevated area, the water table varies from a couple of ten meters depth to close to the surface water as the boundaries (especially around the river). We address this limitation in our discussion, Line 517 – 521.

Comment #8: line 168: "in each period”

Response #8: Thank you for your suggestion. We have adjusted the manuscript accordingly.

Comment #9: figure 2 could perhaps be simplified by only showing the workflow once but then with different climate and abstraction forcing

Response #9: Thank you for your suggestion. We like the idea of simplifying the workflow, however, we also want to show the ‘repetitive’ nature of the simulation. By visualizing the workflow ‘twice’, we want to deliver the message that the model during the baseline simulation (the left part of the figure) has been calibrated and even reported in our previous studies, and the current manuscript is the application of the one-way coupled scheme to project the future groundwater availability under various scenario (the right part of the figure). We think that this message is harder to convey should we simplify the figure to just one workflow, therefore suggesting to keep the figure as it is. We hope you understand and are content with our reasoning.

Comment #10: line 181: the method for estimating potential ET is based on temperature and radiation and thus ignores potential changes in humidity and wind - can you justify this simplification or discuss its impact? Also, this method was apparently not developed for computing potential ET, so why is it applicable for this purpose?

Response #10: Thank you for your question. Indeed, considering the data limitation and uncertainties, we chose to involve methods in estimating PET to be simple. The justification is shown in the referred paper (de Bruin, et al. 2016), that the method is mostly confined to cases without local advection, which means that the local evaporation is not influenced as much by the moving moisture. As our study location is located in a ‘closed system’ (the Bandung area is basically a ‘bowl’ surrounded by mountains), the moisture taken away by the wind and humidity difference is insignificant, therefore allowing for the methods application in the area. In the same paper, it was also mentioned that the terminology potential evapotranspiration itself consists of multiple components, and our understanding of the term has evolved from time to time. Depending on the context of the usage, the methods can also be used to estimate the ‘so-called potential evapotranspiration’. There have been a lot of studies that use this method to estimate potential evapotranspiration too (Oudin, et al. 2005, Imhoff, et al. 2020, Gebremehdin, et al. 2022).

Comment #11: line 196: did you check that the surface shortwave radiation from MRI-ESM2-0 is consistent with that from GFDL-ESM4? Or alternatively explain why this comparison is not needed.

Response #11: Thank you for your question. We did check the surface shortwave radiation from MRI-ESM2-0 and compared them with ones from GFDL-ESM4. There are minor differences (obviously), in which the MRI-ESM2-0 projection results in slightly larger changes in future radiation. However, the effect of these differences in the following calculation is insignificant. First, the radiation is one variable among the other variables to compute the potential evapotranspiration (PET). Then, the PET is used not as a direct number/input in a calculation, but as the upper limit of the calculated actual evaporation. The actual evaporation then influences the magnitude of available water to flow as runoff or to infiltrate and further recharge the groundwater. These multi-steps computation process dissipates the influence initiated by the minor differences between the MRI-ESM2-0 and the GFDL-ESM4 projection radiation products.

Comment #12: line 230: are there any rain gauges in the area to check the assumption of treating CHIRPS as ground truth?

Response #12: Thank you for your question. Yes, there are 11 rainfall stations within and around the area. We have discussed and reported our analysis (in our previous study) on the rainfall estimates in the area by comparing numerous rainfall estimation methods: rainfall station measurement, satellite products, ground-corrected remote-sensing products, and interpolated rain gauge-based estimates. While the rainfall estimates measured by the rainfall stations are supposed to be the most accurate, they represent point measurement as opposed to areal estimates. Therefore, to derive the best estimate of the areal rainfall, we applied an uncertainty quantification method (Extended Triple Collocation method) in our previous study and concluded that CHIRPS could be used as the best estimate in representing the ground-truth areal rainfall. We did not provide a lengthy discussion on this matter in our manuscript, but it is shortly mentioned in Lines 264-265, Section 2.4.2: Wflow_sbm model setup.

Comment #13: line 250: river discharge I assume

Response #13: Thank you for your suggestion. We have adjusted the manuscript accordingly.

Comment #14: line 252-253: shouldn't you be using bias-corrected CMIP6 data for the historical period? Similar comment for figure 4: show the bias-corrected MRI-ESM2-0 for historical period instead of CHIRPS.

Response #14: Thank you for your suggestion. We intentionally use CHIRPS as we treat it to represent the ground-truth rainfall estimates. We also treat the following Figure 5 similarly: using ERA5 data as the ground-truth instead of bias-corrected MRI-ESM2-0 estimates. We would like to keep the consistency in all the figures, not only among these figures but also with Figure 2 (the one about the workflow mentioned above). In that figure, the workflow is ‘repeated’ between the left and the right part of the figure. The left part represents the simulation using the ‘ground-truth’ as the forcing and boundaries, while the right part represents the future projection. We also do not directly compare the ‘ground-truth’ with future projections in Figures 4 and 5, but apply the bias-correction beforehand, with the aim that these comparisons are referenced on the more similar (bias-corrected) yet consistent (with the previous figures) ground.

Comment #15: line 274: what are the values for the storage parameters? and what are the "river-related parameters? Do the latter overlap with parameters in the wflow model?

Response #15: Thank you for your question. The storage parameter includes the specific yield (s_y) and specific storage (s_s) of the unconfined and confined aquifer. The s_y value is 0.2, and s_s 8.7*10^-3 /m. Again, this value has been reported in our previous studies referred to in the manuscript. The river-related parameters include river cell, riverbed hydraulic conductance, surface water elevation (as the head boundary to the groundwater), and the elevation of the bottom of the riverbed. We use different values as the cell size in the surface hydrology and groundwater flow models are different, but they represent similar conditions. For example, in the groundwater flow model, we could not specify the river width, as cells assigned as rivers have to have the whole cell treated as a stream. In reality, the cell width of 100 m is not always consistent with the actual river width, thus we adjust our parameterization in order to make them mathematically representative.

Comment #16: figure 5b: not clear what the extra horizontal lines are in this plot

Response #16: Thank you for your detailed review. The one in the boxplot is the surface downwelling shortwave radiation, while the one at the top (also in the boxplot, but very thin that they look like lines) is the top of atmosphere incident shortwave radiation. We have revised the figure accordingly to make it clear by adding a dashed line, separating the two mentioned ‘regions’ (the revised figure is also attached as a supplement to this response).

Comment #17: line 374: and much smaller storage coefficient in the confined aquifer?

Response #17: Thank you for your suggestion. We agree with your analysis on this matter, thus we have revised the manuscript accordingly.

Comment #18: figure 7: make colorbar title and labels more readable.

Response #18: Thank you for your suggestion. We have revised the now Figure 8 (there is an additional Figure 7 for the hydrological variables accompanying the groundwater recharge) accordingly by enlarging the size of the colorbar title and labels. The revised Figure 8 is also available as a supplement to this response.

Comment #19: line 416: do your simulations predict decreases in baseflow?

Response #19: We did not initially collect the baseflow simulation output intentionally. Thanks to the input from your previous suggestions of accompanying the groundwater recharge with other variables, we now have these values in hand. The groundwater level projections show that the least significant future groundwater table changes are found in the unconfined layer along the stream. This shows that the river baseflow is relatively constant. While this seems not to be a problem, the groundwater table in the unconfined storage actually depletes the most in the mountainous region, contributing to conserving the river baseflow due to the groundwater abstraction along the river downstream part. This is now added to the manuscript (shown above in the discussion part after the new figure showing the aquifer water balance components).

Comment #20: line 479: what do you mean by "pseudo water table"?

Response #20: Thank you for your question. As said in that particular section of the manuscript, ‘In the one-way model coupling setup, groundwater recharge is fully controlled by the surface processes and the pseudo-water table’. This sentence refers to the computational scheme of the hydrological model Wflow_sbm, where the soil is considered as a bucket with a certain depth and divided into saturated and unsaturated zones. The top of the saturated store represents the pseudo-water table. The formula and detailed description of all the Wflow_sbm parameters are now available online in this paper: van Verseveld, W. J., Weerts, A. H., Visser, M., Buitink, J., Imhoff, R. O., Boisgontier, H., Bouaziz, L., Eilander, D., Hegnauer, M., ten Velden, C., and Russell, B.: Wflow_sbm v0.7.3, a spatially distributed hydrological model: from global data to local applications, Geosci. Model Dev., 17, 3199–3234, https://doi.org/10.5194/gmd-17-3199-2024, 2024.

Comment #21: line 485: "In regions with higher margins between the groundwater recharge and soil capacity". Not clear, please clarify.

Response #21: Thank you for your question. In that section, we aimed to describe the circumstances where changes in climate variables would have a higher influence on the changes in groundwater recharge. According to the result in the test basin, the threshold for the water in the vadose zone to infiltrate as groundwater recharge is generally low; the vadose zone is relatively wet due to many factors: the surface processes, the continuous rainfall period, the soil characteristics, etc. Therefore, in regions with higher soil capacity and lower soil moisture, generally, there is a higher possibility of increasing groundwater recharge due to changes in climate variables. We hope that this explanation provides more clarity.

Comment #22: check erroneous text on line 576.

Response #22: Thank you for your detailed review. We have removed the text.