Multi-level numerical and statistical analysis of the hygrothermal behavior of a non-vegetated green roof in a mediterranean climate

Green Roofs (GR) represent a sustainable technological solution for reducing the environmental footprint of urban areas. Despite their benefits, traditional GRs have been criticized regarding their economic feasibility, suggesting to develop advanced hybrid engineering solutions able to simultaneously maximize their hydrological and energetic benefits. In this view, there is a need of numerical models able to describe their complete hygrothermal behavior. Thus, the main aim of this study was to assess the suitability of the one-dimensional mechanistic model HYDRUS-1D in providing an accurate and comprehensive description of the coupled water-heat-vapor transport in a field-scale Non-Vegetated Green Roof (NVGR) in the south of Italy. A complete calibration framework, which encompassed the Particle Swarm Optimization (PSO) algorithm and the combined Global Sensitivity Analysis-Generalized Likelihood Uncertainty Estimation (GSA-GLUE) method, was used to estimate the substrate thermal properties and assess the model predictive uncertainty. The calibrated model was exploited to examine the cooling efficiency of a combined Stormwater Reuse-NVGR system in the warm season. The analysis revealed that deeper substrates are positively correlated with thermal lag and attenuation, and that the irrigation can be properly designed to trigger the evaporative and convective cooling of the NVGR. The Response Surface methodology was finally used to optimize the watering regime on an 8 cm-deep NVGR. The exploitation of the evaporative cooling effect of the NVGR by means of a model-based irrigation optimization led to a reduction of the average soil bottom temperature of 4 °C. The coupled system was able to maximize the energetic benefits of GR.