Regression modeling of combined sewer overflows to assess system performance

Combined sewer overflows (CSOs) occur when untreated raw sewage mixed with rainwater, runoff, or snowmelt is released during or after a storm in any community with a combined sewer system (CSS). Climate change makes CSOs worse in many locales; as the frequency and severity of wet weather events increases, so does the frequency and volume of CSO events. CSOs pose risks to humans and the environ -ment, and as such, CSS communities are under regulatory pressure to reduce CSOs. Yet, CSS communities lack the tools needed, such as performance indicators, to assess CSS performance. Using the city of Cumberland, Maryland as a case study, we use public data on CSOs and precipitation over a span of 16 years to identify a new critical rainfall intensity threshold that triggers likely CSO incidence, and a multiple linear regression model to predict CSO volume using rainfall event characteristics. Together, this indicator and modeling approach can help CSS communities assess the performance of their CSS over time, especially to evaluate the effectiveness of efforts to reduce CSOs.

Automated summary

Combined sewer overflows (CSOs) occur when untreated raw sewage mixed with rainwater, runoff, or snowmelt is released during or after a storm in any community with a combined sewer system (CSS). This study aims to provide CSS communities with tools to assess the performance of their CSS systems over time, especially in evaluating efforts to reduce CSOs. A new critical rainfall intensity threshold and a multiple linear regression model are identified and used to predict CSO incidence and volume based on rainfall event characteristics.