Scaling invariance behaviour of thermal fluxes from an extensive green roof

Green roofs are widely recognized as a Nature-Based Solution that regulates the air temperature within urban environments. Thanks to the shading effect and the evapotranspiration process (ET), the temperature decreases in the green roof surrounding area. Hence, the implementation of green roofs in urban environment for this purpose requires the quantification of ET-related processes at different scales. Nevertheless, because of complexity of the ET process, different methods of measurement have been used at different scales. However, no agreement about the way to assess ET rates over green roofs has been reached between the scientific community, as well as its behaviour at different urban scales is still unclear. Therefore, more investigations on ET measurements are required for better understand and analyse its spatial and temporal variability at different scales.

For this purpose, a Larger Aperture Scintillometer (LAS) MKI from Kipp&Zonen was installed over a wavy-green roof of 1 ha, the Blue Green Wave (BGW), located in the Ecole des Points ParisTech (France). The main objective of this set-up was to assess the refractive index-structure parameter (C_n²) fluctuations from which ET can be deduced by means of the Monin-Obukhov similarity theory and the surface energy balance. As LAS is mainly influenced by fluctuations of air temperature, a radiometer equipped with a temperature sensor was installed in addition over the BGW. Then, the scaling statistics of C_n² and temperature were studied through their power spectral density and their structure function.

The results obtained from the power spectral density demonstrated the scaling invariance of C_n² and temperature over certain ranges of scales. The spectral exponents are close to 5/3 for C_n² and to 2 for the temperature. Regarding the scaling exponents of the structure functions, the multifractal feature of the structure parameter C_n² and the temperature was confirmed. The scale-invariant properties of the empirical data were characterised using the Universal Multifractal framework.