General comments

Good research in the field of groundwater hydrology that has been approached with a worldwide angle. However, important detail is missing. Please, take into account my minor points to fix the issues.

Specific comments

Line 18. “In recent years, much effort has been devoted to the analysis of the sensitivity of groundwater systems to climate change”. Add recent literature on the effects of climate change in mountain ranges, the aquifer recharge from the snow is very sensitive to the climate:

- Lorenzi, V., Banzato, F., Barberio, M. D., Goeppert, N., Goldscheider, N., Gori, F., Lacchini A., Manetta M, Medici G, Rusi S, Petitta, M. (2024). Tracking flowpaths in a complex karst system through tracer test and hydrogeochemical monitoring: Implications for groundwater protection (Gran Sasso, Italy). Heliyon, 10(2).

- Langman, J. B., Martin, J., Gaddy, E., Boll, J., & Behrens, D. (2022). Snowpack aging, water isotope evolution, and runoff isotope signals, Palouse Range, Idaho, USA. Hydrology, 9(6), 94.

Line 48. Cleary mention the 3 to 4 specific objectives of your research by using numbers (e.g., i, ii and iii).

Line 78. “l is the pore connectivity”. Your research appears to focus on porous aquifers of siliciclastic nature (plio-quaternary age?). This point is not clear by reading the manuscript.

Lines 80-81.“Sand, silt, and clay contents”. The geological nature of your aquifers have not been disclosed, see also my comment above.

Line 184. “Under some future climate predictions, the frequency of extreme events is expected to double”. Please, be more specific. Are you talking about semi-arid / arid regions?

Line 184. “Under some future climate predictions, the frequency of extreme events is expected to double”. This sentence should be expanded and moved to the discussion section.

Lines 231-240. The conclusion is too short, it needs more detail.

Lines 232-234. “Our results suggest...rainfall statistics”. The sentence is unclear and too long. Please, split it in two parts.

Line 237. “Focused processes”. Which processes? Please, be more specific.

Figures and tables

Figure 1a. You also have study sites and aquifers in highly arid settings, this is not clear in the text. You don’t have only semi-aridity.

Figure 1c. You can also report Mean Error, Mean Absolute Error and RMS in the graph.

The present paper addresses the impact of changes in rainfall distribution and potential evapotranspiration on groundwater recharge (GR) in semi-arid areas. In their introduction based on the scientific literature, the authors note that "no conclusive generic outcomes can be drawn regarding the relationship between changes in climate conditions and the resulting changes in GR rates", "it is unclear whether the climate variability is amplified or smoothed in the GR response" and "even the trend of the GR response is uncertain".  These assessments are reasonable and I fully agree with them.

Several factors may explain this large range of GR responses to present climatic changes, which leads to apparent contradictions in their recent evolution. The first explanation is the variability of environmental conditions in the natural state. The second explanation is the multiplicity of scientific approaches (various methods using various types of data sets monitored at different scales in space and over time), which logically lead to heterogenous results. Moreover, depending on methods, the calculated GR represents an integration over a very variable time. The third explanation is that in the last decades GR may have changed a lot as a consequence of the climate change and the multifaceted human modifications of semi-arid landscapes (e.g. changes in land use and land cover; water conservation works). The GR estimates found in the literature aggregate values from various stages of this evolution between areas still in a mostly natural state and others deeply modified. Depending on areas, direct human intervention may be a much stronger factor than climate change. For instance, the increase in GR by one order of magnitude in southwestern Niger (e.g. Favreau et al., 2009) and by two orders of magnitude in eastern Australia (e.g. Allison et al., 1990) was explained by a change in the vegetation cover only. Therefore raw data from the literature should be used with a cautious reference to their specific contexts, which is not really the case in the submitted text.

An important concern is the geographical extent of this work. The authors used GR estimates from 200 semi-arid locations in different continents, in a wide range of soils and climate conditions. The annual rainfall in the sites considered ranges from 180 to 1044 mm, with more than one half between 400 and 600 mm. In fact, 60 % of the sites are in Australia, 20 % in Africa, 10 % in North America and 10 % in the other continents, which differs significantly from the distribution of semi-arid areas in the world. Does this selection bias the final conclusions? The Figure 1-a will probably surprise many readers who usually see a much larger extent of semi-arid areas in global maps; this singularly restricted coverage should be justified.

Another important concern is the very restrictive assumptions for the calculation: (i) GR occurs only through diffuse recharge (i.e. without any focused recharge); (ii) transpiration is negligible vs. evaporation; (iii) surface runoff is negligible; (iv) there is no preferential flow in the unsaturated zone. The fist assumption contradicts the observation that focused and diffuse recharge often coexist in the same area. Their respective proportions depend on local geomorphological conditions (e.g. Cuthbert at al., 2019). At the global scale, it is generally accepted that the proportion of focused recharge increases with aridity and as a consequence the driest semi-arid regions would be excluded from this calculation. The second assumption neglects transpiration uptake while the vegetation cycle in semi-arid areas is closely linked with the rain distribution, which is also the driver for GR. The third assumption requires to limit the application of the calculation to very flat areas and/or very low rainfall. The fourth assumption requires a very poor biological activity (roots and fauna).

These four assumptions together are so constraining that the geographical extent of the concerned semi-arid areas is probably very small. The practical relevance of this text appears therefore limited and the added value for researchers working on groundwater in semi-arid areas may be seriously questioned. The authors are conscious of these weaknesses and in their conclusion they mention the possibility of extending their work, but this last precaution is not enough to give the text a convincing strength.

Allison et al., 1990. doi.org/10.1016/0022-1694(90)90030-2

Cuthbert et al., 2019. doi.org/10.1038/s41586-019-1441-7

Favreau et al., 2009. doi.org/10.1029/2007WR006785