This research investigates the application of inverse green roof models to both characterize green roof thermophysical properties and simulate the heat flux through the roof assembly. The first chapter of this thesis presents a preliminary analysis of green roof thermal performance in Ottawa, Canada, using two years of measured data, wherein thermal performance is defined as the percent reduction in total monthly heat exchange promoted by the green roof, relative to that through an adjacent conventional roof. Climatic factors influencing thermal performance are discussed on a seasonal basis, concluding that the green roof functioned best during warmer months when evapotranspiration was likely to be greatest. Thermal performance ranged between 31 -63% from May through September, with reduced performance during the colder months. In chapter 2, two inverse models of varying spatial discretization are developed for the same green roof: (1) a resistor-capacitor (RC) thermal network model, and (2) an implicit finite difference (FD) model.Each model was calibrated using monthly data from May to September in 2016 by employing a genetic algorithm to extract the thermophysical properties of the green roof soil and canopy layers via multi-linear regression. The difference in spatial resolution between each model was identified as an influential factor to thermophysical property estimation during calibration. The calibrated models were used to predict hourly rates of heat flux through the structural component of the green roof over each month of the study period, resulting in a root-mean-squared-error between 0.51 -1.0 W/m 2 and 0.41 -0.81 W/m 2 for the RC and FD models, respectively. Both models were validated against 5 continuous months of data from 2017, demonstrating that inverse modelling can successfully generate realistic thermophysical green roof properties.