Soil Carbon Recovery and Hydrological Buffering in Mongolian Rangelands: Local Benefits and Limits of Earth-System Connectivity

Soil organic carbon (SOC) restoration in degraded drylands is increasingly promoted for climate mitigation and hydrological co-benefits, yet the scale at which these benefits remain detectable remains uncertain. We therefore evaluate SOC recovery and its hydrological effects across three scales—local, basin, and Earth-system—for a 220,966 ha rangeland restoration pilot in Öndörshireet soum, Central Mongolia, using a cold-calibrated RothC model, Monte Carlo uncertainty propagation, machine-learning-assisted measurement support, and basin-scale mass-balance accounting. Baseline SOC averaged 28.27 tC ha−1, and 20-year RothC simulations projected SOC gains of 0.9–6.4 tC ha−1 across rotational, fenced, and seeded–fenced strata. Monte Carlo emission-reduction rates were 0.33–2.19 tCO2e ha−1 yr−1, and the conservative VM0042 deduction chain yielded approximately 10,900–12,200 verified carbon units per year. At the local scale, translating SOC gains into plant-available water capacity produced a small but positive hydrological response—reduced flash runoff, lower flood exceedance and storm runoff, higher baseflow contribution, and lower streamflow flashiness—expressed through flow stabilization rather than increased annual water volume. Across the three scales evaluated here, SOC recovery yields a small increase in plant-available water capacity (≈0.09 mm) locally, associated with reduced flash runoff and greater flow stability; at the basin scale, the signal is three to four orders of magnitude below interannual discharge variability; and at the Earth-system scale, no measurable freshwater or global cooling effect is supported. National-scale estimates indicate substantial credible mitigation potential but negligible Arctic freshwater impacts. By separating local mechanisms from basin- and Earth-system detectability, this framework distinguishes credible local co-benefits from Earth-system over-attribution in dryland carbon projects.