Abstract
<jats:p>Karst aquifers are highly heterogeneous, with complex hydraulic interactions between conduit and matrix domains. Although recession analysis is a key tool for characterizing these systems, existing methods typically analyze individual recession events or rely on deterministic, manual fitting of a single Master Recession Curve (MRC), neglecting event-specific variability and parameter uncertainty. We present a novel automated master recession routine based on a two-linear reservoir model that explicitly simulates dynamic matrix-conduit interaction. <p>We introduce an event-dependent parameter, the initial matrix fraction (f), to represent antecedent matrix storage at recession onset. For the first time in karst recession analysis, we apply a Monte Carlo framework with Generalized Likelihood Uncertainty Estimation (GLUE) to quantify recession parameter uncertainty robustly. Testing across six karst springs in diverse climates produced high Kling-Gupta Efficiencies (KGE &gt; 0.70).</p> <p>Results show that calibrated f is strongly correlated with the &ldquo;old water&rdquo; fraction from independent hydrochemical electrical conductivity (EC) data, confirming its value as a physically meaningful proxy for aquifer drainage dynamics. We also introduce the Karstification Index (KI), a standardized metric quantifying hydraulic contrast between domains and providing an objective measure of aquifer functional maturity. The framework addresses the master recession &ldquo;cloud&rdquo; curve construction problem by identifying the transition phase, the period of greatest uncertainty as flow dominance shifts from conduit to matrix. By automating flow-component separation and uncertainty quantification, the method provides an honest representation of subsurface processes and offers a universal diagnostic tool for large-sample hydrological modeling of karst and non-karst multi-porosity systems globally.</jats:p>