| 13 Oct 2025
Influence of groundwater recharge projections on climate-driven subsurface warming: insights from numerical modeling
Abstract. Groundwater warming due to rising surface temperatures has been documented in both urban and natural settings. However, the potential for long-term changes in the magnitude and seasonality of groundwater recharge to modulate this warming trend has not yet been systematically investigated. In this study, we integrate a stochastic weather generator, distributed hydrologic modeling, and regional thermo-hydraulic groundwater modeling into a unified workflow and apply it to the area of Brandenburg (northeastern Germany). We conduct numerical simulations to assess changes in the subsurface thermal field between present day and 2100, evaluating two climate change scenarios, and incorporate a spectrum of ensemble-based and discrete recharge projections. Our results demonstrate that, while surface temperature rise is the primary driver of the projected groundwater warming of up to 2.5 °C, groundwater flow is responsible for its regional variability in magnitude and affected depths. Higher hydraulic gradients on topographic highs and increased thickness of the permeable Quaternary unit may allow the warming signal to propagate below 200 m depth, whereas groundwater discharge in the river valleys tends to limit it to <200 m. By the late century, the difference in groundwater temperatures between recharge-reduction and recharge-increase scenarios can reach 0.4 °C. Under the high-emissions pathway, a 20 % recharge reduction, from a mean of 75 to 60 mm a-1, causes a 2–5 m water level decline, reducing the area of unconfined aquifer subjected to seasonal temperature fluctuations. Model experiments show that even a hypothetical increase in winter recharge does not suffice to counteract the groundwater warming induced by rising surface temperatures. Changes in advection rates are not expected to affect the net climate-driven accumulated heat in the subsurface due to counterbalancing of heat gains and losses between recharge and discharge areas. Nevertheless, long-term reconfiguration of the potentiometric surface may further impact both the annual and long-term thermal state of key aquifers targeted for water supply and shallow geothermal energy utilization.
Tsypin and colleagues have submitted a manuscript on evolving subsurface thermal regimes, with the key novel contribution being the role of changing recharge. They conclude that advection plays a role in general, but that changing advection (e.g. more cold-season recharge) cannot overwhelm the overall pattern of groundwater warming from diffusion. The study is generally well written and interesting.
A few comments, in no particular order:
1. Many times in this manuscript the authors use 'temperature' to refer (I think) to surface air temperature; such examples can often be found when referring to climate data. But in a manuscript focused on groundwater temperature, they should always be clear about where in the earth system domain they are referring to when they mention temperature.
2. The authors miss some quite related studies from the Netherlands on the role of groundwater flux on subsurface warming. For example Bense et al. (10.1029/2019WR026913) show how changes to groundwater flow regimes can be inferred from temperature profiles (a close topic to the present manuscript although this former study did not forecast into the future), and Bense et al. 2017 (10.1002/2017GL076015) showed that recharge influences where the inflection point occurs in evolving temperature profiles (see L542, where I think a conference abstract is cited for this point). Also, Figure 9 of the present study shows deep temperature shifts when the flow regime changes. This was also demonstrated by Bense et al. 2017 (10.1002/ 2017WR021496) who showed that using a linear temperature-depth profile as an initial condition results in deep shifts in temperature with time because that initial condition doesn't have the same groundwater flow regime as the forward modeling (e.g., Taniguchi et al. 1999 solution). Side note: I think the authors could do a better job of explaining that groundwater flow regime shifts are what is driving the temperature changes at depth in some of their results (which could not happen under heat diffusion alone under the timescales considered here).
3. Figure 1b - it is stated that this is a 3D view of the model mesh. However, neither the mesh nor the model parameters are indicated (from what I can see). This is just a geological model (or perhaps geometric model of the model domain). Much is made of the geology in this paper (rightfully so), but I think it would be useful to just indicate the parameters on this model domain (at least k and Ss - or put in a clear table). If a table, the thermal properties for each layer would be useful to list too (in the main text).
4. The recharge modeling is impressive - it might be good to be more explicit that all of these recharge realizations do not translate into equivalent model runs for groundwater temperature. I only figured that out later, but I admit to reading this in a moving car with children yelling in my ear etc. The authors calibrate this to streamflow. How does the model performance compare for just baseflow? That might be more comforting to know, since that is a better indicator of whether the recharge and discharge processes are being captured reasonably.
5. Eq. 1 - units don't work - is Recharge in units of 1/s (like a volume flux per unit volume)?
6. This is a big numerical model domain with many elements. Was this solved with high-performance computing or anything - or just run on a 'regular' computer? My simulations with this many nodes took weeks to run, but that was 10 years ago.
7. Boundary conditions questions: a) is there any vertical hydraulic gradient on the vertical boundary conditions at the sides? b) Does the model allow for exfiltration at the surface if the water table rises too high? c) I'm surprised the surface air temp and ground surface temperature are so similar. Is there no snow here? The mean annual air temp of 9C suggests there would be.
8. Using mesh size for characteristic length for Peclet number is arbitrary (but maybe common) - how do the authors know what the threshold is for this Pe formulation for which advection matters? Later on they present Pe values and use these to highlight the role of advection, but this is not a formulation for which Pe =1 implies that advection = conduction (like say the Bredehoeft and Papadopulos 1965 formulation does - if I remember right). So what is the threshold for when advection matters? Maybe that is hard to say.
9. The recharge change runs in the groundwater model are interesting (see Table 1), but I'm still not clear where some of these came from (esp. the discrete synthetic scenario). I guess at least one is more of a sensitivity run rather than an expected scenario? Also, it would be helpful if Table 1 were more precise about timelines rather than "present" and "late century". What years are the before/after runs here for?
10. Model calibration - consider presenting the RMSE normalized to head range. I think that would do a better job of showing the fit. 6.6 m sounds high, but the head is highly variable, so it is not a bad fit at all.
11. Where do groundwater temp measurements come from for calibration? Case borehole temp profiles? Pumped groundwater with temp recorded? Open boreholes? Large diameter? Are convection or seasonal bias concerns?
12. Figure 7 - it would be interesting to also see changes in discharge presented somehow. The authors talk about changes in recharge (input) and head (storage), but what about output (relevant for some of the reasons groundwater temperature is relevant)
13. L386 "Profile A, located on a glacial plateau, has a shallow seasonal envelope, entirely above the water table, suggesting no advective transfer of the seasonal thermal signal into the saturated zone" - I don't think you have to be saturated for heat advection to matter. It depends on the water flux, not the saturation per se. if you have recharge through the vadose zone, advection can still theoretically matter.
Minor comments
L60-65 - I think the controversy over the role of advection is overstated here. In the Benz et al. study, the authors used a parsimonious approach in a global study and justified this using 2D numerical models of advection and conduction. They basically found that for typical hydrogeological systems and recharge rates, heat advection is not a primary driver of groundwater temperature change. However, they note that is not the case for some basins. Other studies cited here are in steeper topography or with much higher recharge, and so it makes sense that they found that advection mattered.
Figure 11 a - label missing for scale bar?
Conclusions _ 'depth limit of advection overprint' sounds a bit jargony for the conclusions
In general, these comments are all easy to address I think, and I look forward to seeing this article published.