Quantifying Ecohydrological Responses to Large‐Scale Vegetation Restoration by Incorporating Dynamic Root Processes

Vegetation restoration (VR) can affect ecohydrological processes in water‐limited regions. Paired numerical experiments with and without VR are commonly used to quantify these effects. However, existing studies often rely on static root schemes that cannot capture plant root adjustments under water stress and assume fixed vegetation states in the no‐restoration case, even though climate variability alone can drive vegetation change. To address these limitations, this study applies the Noah‐MP model with a prognostic dynamic vegetation scheme and a dynamic root module to the Chinese Loess Plateau—one of the world's largest VR hotspots. This approach improves the paired numerical experiments' ability to isolate VR effects from climate variability. Relative to the static scheme, the dynamic root configuration improved simulated evapotranspiration (ET), gross primary production (GPP), and leaf area index, with performance gains of 43.8%, 51.5%, and 59.5%, respectively, and the largest improvements occurring in dry years. Moreover, VR increased GPP and ET by 40.4% and 3.9%, and reshaped ET partitioning: transpiration ( T ) increased by 40.4%, while soil evaporation decreased by 19.9%, elevating T /ET by 41.5% and water‐use efficiency by 38.2%. Streamflow declined by 33.9%, driven primarily by reductions in baseflow, which accounted for 60.5% of the total decrease. Land‐use conversions from grassland or cropland to forest produced the most pronounced ecohydrological responses. Overall, VR enhanced carbon uptake but also increased water consumption, highlighting trade‐offs between ecological benefits and water availability. These findings underscore the importance of representing dynamic root processes when evaluating restoration impacts and supporting sustainable restoration planning.