Development of multi-metamodels to support surface water quality management and decision making

Watershed management and planning is a complex decision-making process, which not only involves deliberation using one or more watershed models, but also requires collaboration among multiple stakeholder groups with different ideologies, interests, and demographics. Web-based decision support tools have great potentials to enhance the transparency and participation of such decision making processes. Although physically based surface water quality models are well suited for offline water quality analyses, they are often too computationally demanding to be deployed in a web-based environment. In this work, three metamodels are developed to support decision-making activities related to surface water quality management at Arroyo Colorado Watershed, a coastal watershed located in Texas, US. All three metamodels are trained using an existing Soil and Water Assessment Tool (SWAT) model developed for the watershed. The main objectives of the metamodels are to support web-based decision support, including near-term nutrient load forecasting, online sensitivity study, and long-term load reduction planning. All metamodels either replicate or extend the capabilities of the original SWAT model and, thus, provide proxies for regulators and stakeholders to examine and discuss model results interactively. The novel, multi-metamodel methodology taken here is not only useful for supporting multigroup decision making and public education, but also provides a more effective way to leverage existing investment on watershed models.