Although it is widely accepted that the detention performance of green roofs is of interest to stormwater engineers and planners, no single metric allows detention to be unambiguously defined. Detention effects are highly sensitive to rainfall characteristics and antecedent conditions, and individual roofs typically exhibit wide variations in detention performance between storm events. This paper uses a straightforward hydrological model to explore two alternative approaches to describing detention performance: a probabilistic approach based on long time-series simulations; and a design storm approach. It is argued that the non-linear reservoir routing parameters (scale, k and exponent, n) provide fundamental descriptors of the detention process, with modelling enabling performance to be determined for specific rainfall inputs. The study utilises 30-year rainfall time-series predictions for four contrasting UK locations to demonstrate the utility of the two proposed design approaches and to comment on locational variations in detention performance.

Green roofs may make an important contribution to urban stormwater management. Rainfall-runoff models are required to evaluate green roof responses to specific rainfall inputs. The roof's hydrological response is a function of its configuration, with the substrate – or growing media – providing both retention and detention of rainfall. The objective of the research described here is to quantify the detention effects due to green roof substrates, and to propose a suitable hydrological modelling approach. Laboratory results from experimental detention tests on green roof substrates are presented. It is shown that detention increases with substrate depth and as a result of increasing substrate organic content. Model structures based on reservoir routing are evaluated, and it is found that a one-parameter reservoir routing model coupled with a parameter that describes the delay to start of runoff best fits the observed data. Preliminary findings support the hypothesis that the reservoir routing parameter values can be defined from the substrate's physical characteristics.

A rainfall simulator of length 5 m and width 1 m was used to supply constant intensity and largely spatially uniform water inflow events to 100 different configurations of commercially available green roof drainage layer and protection mat. The runoff from each inflow event was collected and sampled at one-second intervals. Time-series runoff responses were subsequently produced for each of the tested configurations, using the average response of three repeat tests. Runoff models, based on storage routing (dS/dt = I–Q) and a power-law relationship between storage and runoff (Q = kSn), and incorporating a delay parameter, were created. The parameters k, n and delay were optimized to best fit each of the runoff responses individually. The range and pattern of optimized parameter values was analysed with respect to roof and event configuration. An analysis was performed to determine the sensitivity of the shape of the runoff profile to changes in parameter values. There appears to be potential to consolidate values of n by roof slope and drainage component material.

Green roofs have been adopted in urban drainage systems to control the total quantity and volumetric flow rate of runoff. Modern green roof designs are multi-layered, their main components being vegetation, substrate and, in almost all cases, a separate drainage layer. Most current hydrological models of green roofs combine the modelling of the separate layers into a single process; these models have limited predictive capability for roofs not sharing the same design. An adaptable, generic, two-stage model for a system consisting of a granular substrate over a hard plastic ‘egg box’-style drainage layer and fibrous protection mat is presented. The substrate and drainage layer/protection mat are modelled separately by previously verified sub-models. Controlled storm events are applied to a green roof system in a rainfall simulator. The time-series modelled runoff is compared to the monitored runoff for each storm event. The modelled runoff profiles are accurate (mean Rt2 = 0.971), but further characterization of the substrate component is required for the model to be generically applicable to other roof configurations with different substrate.

Flooding represents around 32% of total disasters in Indonesia and disproportionately affects the poorest of communities. The objective of this study was to determine significant statistical differences, in terms of river catchment characteristics, between regions in West Java that reported suffering from flood disasters and those that did not. Catchment characteristics considered included various statistical measures of topography, land-use, soil-type, meteorology and river flow rates. West Java comprises 154 level 9 HydroSHEDS sub-basin regions. We split these regions into those where flood disasters were reported and those where they were not, for the period of 2009 to 2013. Rainfall statistics were derived using the CHIRPS gridded precipitation data package. Statistical estimates of river flow rates, applicable to ungauged catchments, were derived from regionalisation relationships obtained by stepwise linear regression with river flow data from 70 West Javanese gauging stations. We used Kolmogorov–Smirnov tests to identify catchment characteristics that exhibit significant statistical differences between the two sets of regions. Median annual maximum river flow rate (AMRFR) was found to be positively correlated with plantation cover. Reducing plantation land cover from 20 to 10% was found to lead to a modelled 38% reduction in median AMRFR. AMRFR with return periods greater than 10 years were found to be negatively correlated with wetland farming land cover, suggesting that rice paddies play an important role in attenuating extreme river flow events. Nevertheless, the Kolmogorov–Smirnov tests revealed that built land cover is the most important factor defining whether or not an area is likely to report flood disasters in West Java. This is presumably because the more built land cover, the more people available to experience and report flood disasters. Our findings also suggest that more research is needed to understand the important role of plantation cover in aggravating median annual maximum river flow rates and wetland farming cover in mitigating extreme river flow events.