While there are many approaches to stormwater treatment system design, many fail to address the key point:
The majority of stormwater pollutants are found in small, everyday events, hence the systems need to be optimized for those events.
For systems such as raingardens, where treatments occurs continuously during a rainfall event, assumptions on such things as antecedent conditions, rainfall intensity and drying times becomes critical. These things can only be accurately modelled using real historical rainfall data.
Assessment of historical data may provide information on what depth of rain falls in a typical rainfall event, and what size rainfall event one need to capture and retain in order to treat a certain percentage of the annual precipitation. However, assumption as to how much time can pass between two bursts of rainfall while still be considered part of the same rainfall event (usually anywhere between 2 to 12 hours) will need to be made. For systems such as raingardens where the system typically will empty in a few hours, this becomes critical. Designing based on rainfall depth alone is then likely to result in oversized systems.
To accurately design complex systems such as raingardens, continous modelling with historical rainfall data is needed.
Given that the majority of stormwater runoff occurs in small, every-day rain events, sizing systems to accommodate large storm events will only provide a modest increase in annual treated runoff volume compared to smaller systems, with poor return on investment.