A Two-Stage Runoff Detention Model for a Green Roof

Urbanization has caused an increase in per-event stormwater runoff volumes. Existing combined
sewer systems are becoming less able to take in storm runoff without overflowing, which may
cause flooding and water quality issues. Sustainable drainage systems (SUDS) are structures and
practices intended to reduce the volume and rate of a site’s runoff to pre-development levels.
Green roofs, not requiring exclusive land use, can be easily integrated into dense urban areas.
However, their hydrological behaviour requires further understanding.
A generic tool was created for routing detained rainwater through separately-modelled substrate
and drainage layer components of a green roof. Components were monitored in isolation, in
purpose-built rainfall simulators, under laboratory conditions. Configuration variables (e.g. roof
slope) were varied and their effects on runoff response assessed. Nonlinear storage routing
methods were used to fit modelled to monitored runoff profiles, by optimizing routing
parameters. The sensitivity of these parameters to test variables was assessed, greatly reducing
the number of individual values required for modelling either layer.
The runoff response of a two-layered green roof system at field capacity was tested under
laboratory conditions. The substrate model, in series with the drainage layer model, was
parameterized for the two-layered system, and time-series runoff predictions and observations
were compared. The model produced consistently accurate results. This model was reparameterized
for three monitored test beds in Sheffield, UK, using estimated parameter values
for the three untested system configurations. The model was found to be fit for purpose,
approaching laboratory accuracy in the best cases. Peak flow predictions were improved by
allowing limited runoff to occur before a roof’s water content completely reached field capacity.
Further work should extend the model’s applicability to long time-series, through improved
evapotranspiration modelling. Further laboratory observations of individual roof components are
desirable, to increase the range of modellable green roof configurations.