The role of Evapotranspiration in the Hydrologic Performance of Urban Nature-Based Solutions

Green roofs (GR) are widely implemented as nature-based solutions (NBS) to mitigate urban stormwater impacts through rainfall retention, delayed drainage, and enhanced evapotranspiration (ET). However, predicting their hydrologic performance remains challenging because GR operates under highly heterogeneous urban microclimates exhibit strong soil moisture constraints due to shallow engineered substrates, and are often forced using meteorological data that may not represent roof-level conditions. This thesis advances the understanding and modelling of ET and soil moisture dynamics in GR by integrating continuous hydrologic simulation under current and future climates, data-driven analysis of soil water content (SWC), and a systematic sensitivity assessment of reference evapotranspiration (ET₀) to climate inputs. As a foundation, a comprehensive critical review synthesizes existing approaches for estimating potential and actual ET (AET) in urban NBS, identifying fundamental gaps related to scale mismatch, data requirements, and the limited consideration of urban microclimates in conventional ET formulations. This review establishes the theoretical foundation and motivates the need for physically consistent, urban-adapted modelling approaches. First, the present thesis applies continuous long-term hydrologic simulations to quantify GR hydrologic performance under baseline conditions and projected future climate scenarios. Model outputs are used to evaluate key performance indicators such as volume reduction, peak flow attenuation, storage dynamics, climate variability, and to assess how shifts in rainfall regime and atmospheric demand may alter retention capacity and drainage response. Results demonstrate that GR hydrologic performance is strongly event dependent and future climatic changes can modify the balance between rainfall inputs, soil water storage, and drainage generation, with implications for design robustness and climate adaptation planning. Second, the present thesis develops a data-driven framework to predict AET by observing SWC variability in GR using high resolution moisture content level and site-scale monitoring. The analysis characterizes temporal dynamics across wet and dry periods, identifies transitions between energy-limited and moisture-limited ET regimes, and quantifies how substrate water availability governs AET magnitude and persistence. Findings highlight that SWC constraints ET and rapid drying cycles drive pronounced diurnal-to-seasonal AET variability and contribute substantially to uncertainty in long-term water balance closure, especially in shallow substrates. Third, the thesis investigates the transferability of meteorological forcing from standard weather stations to green urban infrastructures by quantifying the sensitivity of ET₀ to climate input variables. Microclimatic variability analyses demonstrate that net radiation, temperature and vapor pressure deficit typically dominate ET₀ variability, while humidity and wind speed effects are strongly site- and season-dependent. The results provide a robust basis for understanding why ET estimates derived from distant or non-representative stations can be biased in urban NBS settings and identify which microclimatic measurements are most critical to reduce uncertainty when local observations are unavailable. Overall, this thesis provides an integrated and urban-relevant framework for improving ET representation and hydrologic performance prediction in GR, supporting more robust design, monitoring strategies, and climate-resilient planning of urban green infrastructure.