Water Quality Model Calibration via a Full-Factorial Analysis of Algal Growth Kinetic Parameters

James D Bowen,Noyes B Harrigan

doi:10.3390/jmse6040137

James D Bowen, Noyes B Harrigan

Open Access

https://doi.org/10.3390/jmse6040137

Copy DOI

Abstract

The two-dimensional, laterally-averaged mechanistic eutrophication model CE-QUAL-W2 version 3.72 was used to predict chlorophyll-a concentrations across two different time periods in the Neuse River Estuary, North Carolina. Chlorophyll calibration was performed for two time periods simultaneously by performing a full-factorial experiment that tested seven algal kinetic growth parameters over three levels for a single algal group. A cluster of up to six computers each running between two and ten instances of the program was used to complete and manage the data for 2187 runs for each period. Six numeric criteria were used to determine which runs performed acceptably, yielding a group of 27 cases that met all of the criteria. Calibration performance of the set of cases outperformed a previously calibrated model using three algal groups that met only four of the six selection criteria. Calibration performed this way allowed for a more rational specification of model calibration performance and provided uncertainty estimates of model predictions, albeit at the cost of a considerable increase in computational requirements that necessitated the use of a computer cluster.

Highlights

Algal blooms have plagued the Neuse River Estuary (NRE) for decades [1,2] and multiple models have been employed to study eutrophication in the estuary [3,4,5]
Inorganic N loading to the estuary has been reduced by 15–25% since the 1990s, this reduction has been offset by an increase in organic N loading of approximately 15% and an increase in N loading from the Trent River tributary of approximately 30% [7]
This article reexamines the procedure used to calibrate the algal constituents in a process-based multi-dimensional water quality model of the Neuse River Estuary that has been further developed to simulate a second, more recent multi-year period

Summary

Introduction

Algal blooms have plagued the Neuse River Estuary (NRE) for decades [1,2] and multiple models have been employed to study eutrophication in the estuary [3,4,5]. Inorganic N loading to the estuary has been reduced by 15–25% since the 1990s, this reduction has been offset by an increase in organic N loading of approximately 15% and an increase in N loading from the Trent River tributary of approximately 30% [7] These changes and the continued presence of blooms and water quality criteria violations point to the need to better understand trends in water quality in the Neuse Estuary that have been observed since the model-based TMDL was performed more than ten years ago. This article reexamines the procedure used to calibrate the algal constituents in a process-based multi-dimensional water quality model of the Neuse River Estuary that has been further developed to simulate a second, more recent multi-year period. To make the large number of runs needed to independently vary the parameters, a program was written that would allow multiple computers to work in parallel to perform many model runs and individually report their progress to a cloud-based file that was used so that all model runs were completed by the computer cluster

Model Set-Up

Algal Growth Parameters

Findings

Summary