Author(s): Elijah Richards; Branko Kerkez

Linked Author(s):

Keywords: Dynamical systems; Flooding; IoT; Nature based solutions; Sensing

Abstract: Nature Based Solutions are increasingly deployed for urban flood mitigation, yet their field performance remains poorly characterized. Most assessments rely on design specifications rather than continuous monitoring of actual hydrological behavior. We address this gap using a network of 11 wireless sensors deployed across rain gardens in three southern Michigan municipalities, capturing subsurface water dynamics at 5-minute resolution over multiple storm events. A dynamical systems approach was used to learn governing equations directly from the data, enabling extraction of physically meaningful drawdown coefficients. Results reveal substantial variability in drainage behavior: some sites exhibited exponential decay consistent with unsaturated flow, while others showed nonlinear dynamics indicative of water table interaction. Notably, residential installations performed comparably to municipal sites despite differences in maintenance regimes, suggesting that design specifications and upkeep alone do not predict performance. These findings highlight the need for site-specific monitoring frameworks that account for subsurface heterogeneity when evaluating nature based solution efficacy.

DOI:

Year: 2026