The Architecture of Damage Nucleation: Multivariate Phenomenological Description, Identification, and Simulation

In most contaminant transport simulations and theoretical approaches, contaminant source zones are conceptualized as volumes with simplistic, known geometries. This contradicts the fact that source zones are the outcome of complex and non-linear physical processes (e.g., DNAPL infiltration), which lead to complex source architecture. The consequences of this assumption propagate as severe errors though all subsequent modelling steps, including the interpolation or geostatistical inversion of down-gradient concentration measurements, and the estimation of contaminant total flux and natural attenuation.

The goal of this project is to assess and overcome the consequences of assumed simplistic source architectures. We will conceptualize source zones as random space functions that can capture complex spatial patterns. Individual objectives are stochastic phenomenological description of source architecture, stochastic inference of architectures and their characteristics from local small-scale and other large-scale data, the search for optimal sampling schemes, random generation of architectures and improved models for equivalent upscaled sources.