Using Logistic Regression to Model the Risk of Sewer Overflows Triggered by Compound Flooding with Application to Sea Level Rise

Document Type


Publication Date



Sewer overflow, Logistic regression, Sea level rise, Coastal resilience, Compound flooding

Digital Object Identifier (DOI)



Coastal wastewater and storm water systems can be overwhelmed during high precipitation events, particularly when compounded by high storm surge that blocks spillways and drainage ways. Sea level rise (SLR) brings increased risk of such compound flooding events, triggering sanitary sewer overflows (SSO) which release waste water into the local environment. A logistic regression model was developed to better predict this risk in southern Pinellas County, FL. Model variables were selected from 2000 to 2017 cumulative precipitation and coastal water levels using both objective and subjective criteria. The 2 day (P2) and 90 day (P90) cumulative precipitation, and 2 day water level maximum (W2) were identified as significant predictors from the p-value of their model coefficients, but required an interaction term P2*W2 for model fidelity. The model correctly hindcasted all 6 identified SSOs from 2000 to 2017. SLR was represented by a range of values up to 0.5 m added to W2. For a SLR of 0.5 m the number of SSO days increased by a factor of 42–52 and the number of individual events increased by a factor of ~15. Subtracting recent SLR from W2 reduced the probability of some recent events, suggesting that SLR already is increasing the rate of SSOs.

Was this content written or created while at USF?


Citation / Publisher Attribution

Urban Climate, v. 35, art. 100752