Quantifying evapotranspiration fluxes on green roofs: A comparative analysis of observational methods

Evapotranspiration (ET) is an important process in green stormwater infrastructure (GSI) aiming to reduce urban drainage, to promote cooling and/or to contribute to an urban hydrological balance restoration closer to the natural one. However, on these structures and particularly on green roofs (GR), its evaluation remains challenging and subject to discussion. Estimates of ET by water balance, energy balance, and an ET chamber were performed on five different plots of a full-scale experimental green roof in Trappes (France). Compared to both water balance (90th percentile range of daily ET values: 0.8 mm/d to 3 mm/d) and chamber methods (90th percentile range of daily ET values: 1 mm/d to 1.5 mm/d), the energy balance (90th percentile range of daily ET values is between 1.8 mm and 3.7 mm) produces higher values, 1 to 2 times higher in cumulative values during common periods. The chamber ET displays a similar trend to the energy balance on an hourly basis, and its values remain within the same range as the water balance evaluations on a daily time-step. All three assessments consistently fell below the potential ET values estimated with the Penman-Monteith formula. Critical issues in ET estimation through experimentation have arisen. Sensible heat flux (H) significantly increases ET values when using the energy balance approach compared to the other two methods. The Water Balance method is practical, but on days following rainfall events, the Chamber method may prove more reliable, albeit more time and labour-intensive. The three methods indicated that the substrate thickness was the main contributing factor to increase ET, with well-maintained herbaceous plants providing higher ET values than sedums in thick (15 cm) substrates. In addition, the substrate's nature, especially its organic content, is another factor that promotes ET in green roofs.