Water storage levels and water storage capacity of an extensive green roof quantified from multi-year eddy covariance measurements

Green roofs provide several ecosystem services that may aid cities in locally adapting to climate change such as the regulation of local air temperatures by evaporative cooling and the limitation of stormwater damage by retention of precipitation water in the green roof substrate. In the past, water storage levels in green roofs have often been inferred from substrate moisture measurements. Here, we test the applicability of recession analysis to quantify water storage levels from the temporal decrease in evapotranspiration during dry periods using latent heat flux densities measured by the eddy covariance (EC) method over the time period of 2015-2020. We found water storage levels to vary between 0.1 and 35.8 mm (median of 4.2 mm). The water storage capacity of 35.8 mm was larger by a factor of >= 27 compared to modelled values for paved urban surfaces (1-1.3 mm). Seasonal variation of water storage levels inferred by EC was characterised by an energy-limited evapotranspiration regime in winter and water limitation during summer. The increase in the green roof vegetation coverage over time resulted in a slight increase in the capacity of the green roof to store water. Water storage levels calculated from in-situ substrate moisture sensors found very similar results compared to the EC recession analysis. Multiyear eddy covariance observations prove a useful tool to quantify and monitor variation of water storage levels in an extensive green roof, as long as evapotranspiration is not limited by available energy.