Adaptive Data Assimilation Framework for Robust Flow Estimation in Gravity-Driven Sanitary Collection Systems

Accurate discharge estimation in gravity-driven sanitary collection systems is vital for urban water management, enabling capacity assessment, development planning, inflow/infiltration (I/I) detection, and overflow prevention. Traditional methods compute discharge naively as the product of cross-sectional area (derived from depth) and velocity, amplifying errors from sensor drift or degradation. We present a refined framework that integrates independent depth and velocity measurements with a depth-variable Manning hydraulic relationship using a nonlinear Ensemble Kalman Filter (EnKF). This approach treats discharge, hydraulic states, sensor biases, and base roughness as latent variables evolving over time, enforcing Manning physics as soft constraints. Key innovations include: (1) depth-dependent Manning roughness via a user-specified variability factor, (2) innovation-based adaptive weighting to detect and down-weight faulty sensors (e.g., velocity drops due to low-flow or fouling), (3) strengthened process model persistence for smooth flow transitions, and (4) enhanced diagnostics for regime switching and outlier rejection. By modeling sensor biases and leveraging adaptive mechanisms, the framework ensures robust discharge estimates even during sensor degradation, shifting reliance toward reliable measurements and hydraulic constraints. Applied to sanitary collection system networks, this method improves operational decision-making by providing continuous, smooth, and reliable flow estimates, addressing challenges posed by time-evolving sensor errors and complex flow dynamics.