Statistic Outlook of Green Roof Stormwater Pollutant Loadings

J.Y. Chen, B.J. Adams and James Li

ABSTRACT

Green roofs are recognised as an effective means for runoff volume reduction and peak discharge attenuation from the perspective of stormwater quantity control; from the perspective of runoff quality, a properly designed and constructed green roof may enhance stormwater quality with reduced pollutant loads to receiving waters. In this study, with on-site data monitored at the rooftop, green roof stormwater is analyzed from both perspectives of runoff quantity and quality in an attempt to characterise relationships between the rainfall (or runoff) and pollutant event mean concentrations (EMCs) along with pollutant loads. By analyzing a total of 12 major pollutants found in green roof runoff, it is recognised that the correlation between the loads of two pollutants appears to be stronger than the correlation between the EMCs of two pollutants. As one of the most common pollutants, the correlation between suspended solids and other pollutants are evaluated for the reliability of using the loads of suspended solids as a surrogate to predict other pollutant loads. Further statistic analysis of pollutant loads reveals that the CDFs of log-normal distribution can fit the observed data reasonably well compared with the CDFs of normal and exponential distributions. The normal distribution tends to significantly overestimate the CDFs of pollutant loads for a given value of the CDF. From the extrapolation of the exceedance probability of pollutant loads based on statistics of sample data, it is realized that log-Pearson distribution is capable of providing close estimates of pollutant loads to the estimates from the log-normal distribution for a given return period. By comparison, Pearson distribution is likely to significantly underestimate pollutant loads with reference to the estimates of log-normal or log-Pearson distributions.


Permanent link: