عنوان

Development of a Behaviour-Pattern Model Testing Protocol to Facilitate Coupled Physical-Socio-Economic Modeling

Number

TL58007

.Language of Text, Soundtrack etc

انگلیسی

Title Proper

Development of a Behaviour-Pattern Model Testing Protocol to Facilitate Coupled Physical-Socio-Economic Modeling

General Material Designation

[Thesis]

First Statement of Responsibility

Peng, Xingyu

Subsequent Statement of Responsibility

Adamowski, Jan

Name of Publisher, Distributor, etc.

McGill University (Canada)

Date of Publication, Distribution, etc.

2020

Text of Note

184 p.

Dissertation or thesis details and type of degree

M.Sc.

Body granting the degree

McGill University (Canada)

Text preceding or following the note

2020

Text of Note

Dynamically Coupled Socioeconomic system dynamics models integrated with physically-based Environmental Models (DCSEM) can aid in resolving complex environmental problems involving dynamic interactions between socioeconomic and biophysical systems. DCSEMs are promising tools for environmental management decisions involving the integration of various viewpoints, disciplines and processes. However, the application of DCSEMs has been limited, mainly due to high parameter uncertainty, interdependent spatiotemporal scales among physical parameters from physically-distributed model components, and the possibility of behaviour-oriented model outputs (as opposed to numerical point outputs) from structure-based system dynamics (SD) model components. Therefore, the application of most conventional (numerical) process-based model testing approaches (i.e., analysis of point values within modelled results) is limited, as these approaches are not applicable to the evaluation of DCSEMs. To address the lack of well-defined model testing approaches for DCSEMs that contain a behaviour-oriented SD model component, this research develops a behaviour-pattern (focusing on general behavioural trends rather than on numerical point values of model outputs) model testing protocol that considers parameter uncertainty, model spatiotemporal scales, and output behaviour patterns, which is suitable for DCSEM testing. The behaviour-pattern model testing protocol is developed, tested, and applied to an existing coupled DCSEM, to simulate water table depth at a study basin in Rechna Doab, Pakistan. In this case study, the behaviour-pattern model testing protocol illustrated a better performance compared to the conventional numerical model testing protocol in terms of adaptability and suitability of the results for DCSEMs. The proposed behaviour-pattern model testing protocol addresses high parameter uncertainty, spatiotemporal scales in the modeling process, and the behaviour-oriented SD model component in three main ways: (1) by selecting and using existing global sensitivity analysis (GSA) methods; (2) by selecting and using existing search algorithms for automated DCSEM calibration and evaluation; and (3) by incorporating existing GSA methods and automatic search algorithms into an existing model coupling tool to ensure ease-of-use. The behaviour-pattern model testing protocol was developed in two stages. In the first stage, a behaviour-pattern GSA procedure was developed to rank and screen a large number of DCSEM parameters to fit output behaviour patterns, and thus reduce the parameter uncertainty and facilitate subsequent model calibration and evaluation. In the second stage, a behaviour-pattern based model testing approach, including DCSEM calibration and evaluation, was developed to calibrate the model based on the fit between output behaviour patterns and observed trends, using the influential DCSEM parameters identified in the first stage. These two stages were then implemented in an existing software environment, and were found to accurately capture spatial heterogeneity within the study basin. The proposed behaviour-pattern model testing protocol considers spatiotemporal scale within the model. It also reduces parameter uncertainty by quantifying model uncertainty using GSA, calibration, and evaluation in a behaviour-oriented approach, and thus improves the application of the model testing approach to DCSEMs. The proposed model testing protocol may serve as a useful tool for implementing more spatially and socioeconomically robust DCSEMs to support decision-making for policy makers and stakeholders, facilitating sustainable environmental management through a holistic understanding of the interactions between socioeconomic and hydrological systems.

Subject Term

Algorithms

Subject Term

Calibration

Subject Term

Hydrology

Subject Term

Markov analysis

Subject Term

Mathematical models

Subject Term

Optimization

Subject Term

Parameter estimation

Subject Term

Sensitivity analysis

Peng, Xingyu

Adamowski, Jan

McGill University (Canada)

Electronic name

p

[Thesis]

276903

a

Y