Global Optimization-Based Calibration Algorithm for a 2D Distributed Hydrologic-Hydrodynamic and Water Quality Model

Hydrodynamic models with rain-on-the-grid capabilities are usually computationally expensive. This makes the use of automatic calibration algorithms hard to apply due to the large number of model runs. However, with the recent advances in parallel processing, computational resources, and increasing high-resolution climatologic and GIS data, high-resolution hydrodynamic models can be used for optimization-based calibration. This paper presents a global optimization-based algorithm to calibrate a fully distributed hydrologic-hydrodynamic and water quality model (HydroPol2D) using observed data (i.e., discharge, or pollutant concentration) as input. The algorithm can find a near-optimal set of parameters to explain observed gauged data. The modeling framework presented here, although applied in a poorly-gauged catchment, can be adapted for catchments with more detailed observations. We applied the algorithm in different cases of the V-Tilted Catchment, the Wooden-Board catchment, and in an existing urban catchment with heterogeneous data. The results of automatic calibration indicate $\mathrm{NSE} = 0.99$ for the V-Tilted catchment, $\mathrm{RMSE} = 830~\mathrm{mgL^{-1}}$ for salt concentration pollutographs (i.e., 8.3% of the event mean concentration), and $\mathrm{NSE} = 0.89$ for the urban catchment case study. This paper also explores the issue of equifinality in modeling calibration (EqMC). Equifinality is defined as the set of different parameter combinations that can provide equally good or accepted results, within the physical parameter ranges. EqMC decreases with the number of events and increases with the choice of partially or nonproducing runoff ones. Furthermore, results indicate that providing more accurate parameter ranges based on a priori knowledge of the catchment is fundamental to reduce the chances of finding a set of parameters with equifinality.