A two-stage approach to evaluate parameter equifinality in build-up/wash-off stormwater quality models

Build-up/wash-off models are widely used in urban stormwater quality modeling to estimate pollutant loads. However, their empirical structure often introduces equifinality, the presence of multiple parameter sets yielding similar predictions, raising concerns about the robustness and physical interpretability of model outputs. In addition to structural limitations, three key modeling considerations – land use representation, pollutant type, and the temporal scale of parameter calibration – can influence equifinality. This study introduces a novel two-stage calibration-evaluation framework that integrates seasonal parameter pooling and sequential conditional/joint likelihood reweighting to examine equifinality while accounting for these factors. The framework is implemented using the Generalized Likelihood Uncertainty Estimation (GLUE) approach and applied to simulate total suspended solids (TSS), total phosphorus (TP), and soluble reactive phosphorus (SRP) loads in a mixed urban land-use catchment in London, ON, Canada. Simulation results indicate that deterministic calibration often yields parameters with limited transferability between events due to parameter equifinality. Land uses with high impervious cover were found to have relatively low equifinality, and seasonal calibration improved parameter generalizability for SRP and TP, but was less effective for TSS. The new proposed framework provides a flexible, performance-driven alternative to deterministic approaches in build-up/wash-off modeling. It enables explicit evaluation of parameter generalizability and predictive reliability under equifinality, supporting robust scenario-based stormwater quality modeling and informed decision-making.