Extensive green roofs can be an important nature-based solution for increasing carbon uptake in cities. However, relatively little is known about the temporal variation of carbon exchange, e.g. due to different weather conditions. In our current study, we used a machine learning method (random forest) to investigate the dependence of carbon exchange on meteorological variables. We were able to draw on a multi-year time series (2015 - 2020) of turbulent CO2 fluxes from Berlin, which we measured using the eddy covariance method on an extensive green roof at BER Airport.
Husting, T., Schröder, B. and Weber, S., 2024. Predicting multi-annual green roof net ecosystem exchange using machine learning. Building and Environment, 263: 111878. https://doi.org/10.1016/j.buildenv.2024.111878
