Automatic Calibration for CE-QUAL-W2 Model Using Improved Global-Best Harmony Search Algorithm

Afshin Shabani,Haochi Zheng,Xiaodong Zhang,Xuefeng Chu

doi:10.3390/w13162308

Abstract

CE-QUAL-W2 is widely used for simulating hydrodynamics and water quality of the aquatic environments. Currently, the model calibration is mainly based on trial and error, and therefore it is subject to the knowledge and experience of users. The Particle Swarm Optimization (PSO) algorithm has been tested for automatic calibration of CE-QUAL-W2, but it has an issue of prematurely converging to a local optimum. In this study, we proposed an Improved Global-Best Harmony Search (IGHS) algorithm to automatically calibrate the CE-QUAL-W2 model to overcome these shortcomings. We tested the performance of the IGHS calibration method by simulating water temperature of Devils Lake, North Dakota, which agreed with field observations with R2 = 0.98, and RMSE = 1.23 and 0.77 °C for calibration (2008–2011) and validation (2011–2016) periods, respectively. The same comparison, but with the PSO-calibrated CE-QUAL-W2 model, produced R2 = 0.98 and Root Mean Squared Error (RMSE) = 1.33 and 0.91 °C. Between the two calibration methods, the CE-QUAL-W2 model calibrated by the IGHS method could lower the RMSE in water temperature simulation by approximately 7–15%.

Highlights

Water temperature is one of the water-quality parameters that is frequently simulated by the CE-QUAL-W2 model, because it is closely related to water mixing and physicochemical and biological processes of lakes and reservoirs [9]
Improved Global-Best Harmony Search (IGHS) and Particle Swarm Optimization (PSO) started their searches by exploring the solution space and converged exploiting parts of the solution space that were detected as the potential location for the global
We found that the uncertainty in PSO calibration was reduced significantly when the length of calibration was increased from 50 to 100 model runs (Figure 8d), but the change was negligible for IGHS (Figure 8c)

Summary

Introduction

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. CE-QUAL-W2 is a hydrodynamic and water-quality model for simulating physical and biogeochemical processes in lakes, rivers, estuaries, and reservoirs [1,2]. The model has been widely used to simulate water balance and water constituents, such as dissolved oxygen, total dissolved and suspended solids, and nutrients in waterbodies [3,4,5,6,7,8]. Water temperature is one of the water-quality parameters that is frequently simulated by the CE-QUAL-W2 model, because it is closely related to water mixing and physicochemical and biological processes of lakes and reservoirs [9]. CE-QUAL-W2 was used to simulate vertical and horizontal profiles of water temperatures in Crystal Spring

Methods

Discussion

Conclusion