Brazil’s Headwater Aquifers Face Steep Declines Under Climate Change

Beneath the forested ridges of southeastern Brazil, an invisible reservoir is quietly shrinking. New research published in Environmental Earth Sciences projects that groundwater storage in two of the state of Minas Gerais’s most important headwater basins will decline dramatically over the rest of the century, with the worst-case scenario showing a staggering 74.4 percent reduction by 2099 in the Grande River basin. The findings, based on detailed hydrological modeling driven by multiple climate models, suggest that even basins that have so far buffered the region against drought may lose that capacity as temperatures rise and rainfall patterns shift.

The study, led by Camila L.C. Ribeiro of the Federal University of Lavras together with colleagues at the University of Illinois Urbana-Champaign and Brazil’s disaster monitoring center CEMADEN, focused on the headwater reaches of the Grande and Jequitinhonha River basins. These two watersheds were deliberately chosen because they sit in climatically contrasting zones of Minas Gerais and have experienced relatively little human disturbance over the past two decades. That makes them natural laboratories for isolating the signal of climate change from the noise of land-use change, a persistent challenge in hydrological impact studies.

The Grande River basin, sheltered by the Mantiqueira Range, feeds one of Brazil’s most significant hydropower systems as the main tributary of the Paraná River. Its climate is humid subtropical, with hot rainy summers and dry winters, and annual precipitation near the range crest can exceed 2,300 millimeters thanks to orographic effects. The Jequitinhonha River basin, by contrast, rises in the Espinhaço Range in north-central Minas Gerais and spans conditions from tropical with a dry winter to hot semiarid, with lower and far more irregular rainfall. One basin is cool and wet; the other is hot and dry. If both are losing groundwater, the researchers reasoned, the implications for the wider region are sobering.

To simulate how these basins will respond to future climate, the team employed the MHD-INPE hydrological model, a physically based distributed rainfall-runoff model developed at Brazil’s National Institute for Space Research. The model represents the movement of water through surface, subsurface, and deep soil layers, and it was run at a two-kilometer spatial resolution for each basin. Because global climate models produce projections at resolutions too coarse for basin-scale analysis, the team used projections from four CMIP5 global models—BESM, CanESM2, HadGEM2-ES, and MIROC5—dynamically downscaled to 20 kilometers by the Eta regional climate model at CPTEC/INPE. The ensemble mean of these projections was then used to force the hydrological model through 2099 under two emissions pathways: the intermediate RCP 4.5, which assumes emissions stabilization by mid-century, and the pessimistic RCP 8.5, which assumes continuously rising emissions and warming of 2.6 to 4.8 degrees Celsius.

A key technical innovation in the study was a stepwise back-correction procedure applied to precipitation data. Sparse rain-gauge networks, particularly in the Jequitinhonha basin with only three stations, make it difficult to capture the true spatial variability of rainfall. The correction algorithm iteratively adjusted precipitation inputs based on discrepancies between simulated and observed streamflow, subject to mass-conservation constraints, and was validated through frequency analysis to ensure the corrected extremes remained within statistically plausible ranges. After this reanalysis, model performance improved substantially, with calibration and validation metrics classified as good to very good in both basins.

The researchers then estimated groundwater storage using a method developed by hydrologist Wilfried Brutsaert, which exploits the tight physical relationship between aquifer storage and the minimum streamflow a river sustains during dry periods. By extracting the annual minimum seven-day moving average of simulated streamflow and relating it to a basin-specific transmissivity parameter derived from baseflow recession analysis, the team reconstructed annual minimum groundwater storage for each scenario and time slice: 2006 to 2040, 2041 to 2070, and 2071 to 2099. Trends were assessed with the Mann-Kendall test and quantified with Sen’s slope estimator at the five percent significance level.

The results were unambiguous. In the Jequitinhonha basin, minimum groundwater storage declined in every future projection compared with the baseline, with the largest reduction of 49.6 percent occurring in the 2071 to 2099 window under RCP 4.5; under RCP 8.5, reductions ranged from 31 to 44 percent. The Grande River basin showed even steeper losses, with the maximum decline of 74.4 percent projected for 2071 to 2099 under RCP 8.5. Significant downward trends appeared in all projections for both basins, and streamflow followed the same pattern, with the strongest reductions concentrated in the second half of the century. For a basin that anchors much of Brazil’s hydroelectric generation, a loss of nearly three-quarters of minimum groundwater storage represents a fundamental threat to dry-season river flows.

Perhaps the most striking insight is why the wetter, more resilient Grande basin suffers the larger absolute decline. The answer lies in the projected precipitation itself: trend analysis showed that rainfall reduction rates were consistently greater in the Grande basin than in the Jequitinhonha across all scenarios and periods, particularly under RCP 8.5 in the later decades. The Jequitinhonha, meanwhile, is inherently vulnerable for different reasons. Nearly half its area is covered by rocky outcrops of quartzite, schist, and metaconglomerate that limit infiltration, and its low-storage, low-transmissivity aquifers make it acutely sensitive to climatic fluctuations. Previous work by the same team had already documented declining groundwater storage trends in the Jequitinhonha of 0.11 millimeters per year over recent decades, driven by climate rather than land use.

The contrast between the two basins highlights the role of geology and soil in hydrological resilience. The Grande basin’s deep, porous Oxisols store substantial water and promote infiltration and recharge, while interactions between Atlantic Forest vegetation, soils, and the Mantiqueira geology enhance preferential flow paths. But resilience has limits. The authors argue that forest restoration and soil conservation measures are urgently needed, especially in pasture and agricultural areas on Cambisols and in groundwater recharge zones, to reduce overland flow and boost infiltration. Native vegetation restoration in strategic areas has been identified in the broader literature as a key adaptation strategy for maintaining hydrological resilience under a warming climate.

The study is not without caveats, which the authors lay out candidly. Simulations relied on the ensemble mean of four global models, so uncertainty across individual models was not quantified. The analysis used CMIP5 projections rather than the newer CMIP6 generation because the high-resolution Eta/CPTEC regionalized datasets were the best fit for headwater modeling, and future work should test whether SSP-based projections alter the conclusions. Sparse gauging, unrepresented water withdrawals, and future land-use change all add uncertainty. Yet the central message stands: significant baseflow reduction is projected throughout the twenty-first century regardless of a basin’s current climate, and the tools exist to act on it. By feeding these projections into water allocation mechanisms and management frameworks, policymakers in Minas Gerais and beyond can begin planning for a future in which the groundwater that sustains rivers through the dry season can no longer be taken for granted.

Subject of Research: Projected impacts of climate change on groundwater storage in headwater basins of Minas Gerais, southeastern Brazil

Article Title: Groundwater storage projection in headwater basins under different climate conditions in Southeast Brazil

Article References: Ribeiro, C. L., Melo, P. A., Guzman, J. A., Chu, M. L., & de Mello, C. R. (2026). Groundwater storage projection in headwater basins under different climate conditions in Southeast Brazil. Environmental Earth Sciences, 85(16), Article 422. https://doi.org/10.1007/s12665-026-13172-0

Image Credits: AI Generated

DOI: 10.1007/s12665-026-13172-0

Keywords: groundwater storage, climate change, hydrological modeling, headwater basins, Minas Gerais, Brazil, baseflow, RCP scenarios, MHD-INPE model, drought, water resources management, Brutsaert method