Decoupling of precipitation and streamflow trends induced by forest expansion: Evidence from time-varying Budyko analysis

56 watersheds distributed across the island of Taiwan. Using a time-varying Budyko framework, this study aims to: 1) quantify shifts in catchment water storage capacity (represented by the landscape parameter, m ); 2) decompose runoff variations into climate-induced and landscape-induced components to identify the drivers of streamflow reduction; and 3) assess the implications of using time-invariant versus time-varying landscape parameters for future water resource projections. This study provides the first quantitative evidence of a widespread “decoupling” between precipitation and streamflow in Taiwan; while 79.6% of watersheds show upward precipitation trends, only 60.7% show upward runoff trends, with many showing significant landscape-driven reductions; 2) Decomposition analysis reveals that enhanced watershed water storage capacity (an increasing m ) has offset climate-driven runoff gains in the majority of watersheds. This increase in storage capacity is strongly correlated with forest expansion, which acts as a primary driver for reduced runoff despite increasing rainfall.; 3) The assumption of stationarity in the Budyko landscape parameter leads to significant biases. Using a time-invariant parameter results in an overestimation of future runoff by up to 240 mm under RCP8.5 scenarios, highlighting the necessity of incorporating time-varying parameters for accurate hydrological projections in changing environments. • Reveal widespread decoupling of precipitation and streamflow trends across Taiwan. • Quantify shifts in water storage capacity via a time-varying Budyko framework. • Enhanced water storage capacity offsets climate-driven runoff gains. • Forest expansion is a primary driver of landscape-induced runoff reduction. • A plausible estimate of future runoff requires time-varying parameterization.