Using Hydrograph Disaggregation to Improve the Sanitary Sewer Hydrology Model Calibration Process

Modeling sanitary sewer flow affected by infiltration and inflow is challenging due to complex interactions among environmental, infrastructure, and operational factors. Despite these challenges, sanitary sewer flow models are essential for developing design hydrographs, generating long-term synthetic flow time series, assessing system adequacy, and evaluating design alternatives. These models are commonly calibrated using short-duration flow monitoring data, which often do not capture the full range of hydrologic conditions of interest.

Continuous sanitary sewer flow models typically simulate multiple processes simultaneously, which are summed to generate a synthetic sewer flow time series. These processes include diurnal variations in residential, commercial, and industrial discharges; short-term inflow responses to rainfall and snowmelt; longer-term infiltration driven by rainfall and snowmelt; and seasonal variations associated with groundwater infiltration. Each process is modeled using calibratable parameters and independent time series, such as rainfall and temperature. Traditional calibration approaches adjust parameters for all processes simultaneously to match observed sewer flow, resulting in large parameter spaces and increased uncertainty.

In this study, we present an alternative calibration approach that first disaggregates the measured sewer flow signal into four components: diurnal variation, fast response, slow response, and long-term seasonal variation. Separate hydrologic models are then independently calibrated to each component, yielding distinct parameter sets for each process. By limiting the number of parameters calibrated at one time and matching each model to a specific sub-signal, this approach significantly reduces the parameter solution domain. As a result, calibration is faster, more reliable, and more accurate, with a substantially smaller parameter joint confidence region.

The effectiveness of this calibration approach was evaluated using rainfall and flow data provided by the Milwaukee Metropolitan Sewerage District (MMSD), a Lynn-Hollick filter to disaggregate the metered flow hydrograph, and the Antecedent Moisture Model (AMM) as the hydrologic modeling framework.