Use of emulation modeling on a process-based hydrological model for spatial land use decision support: proof-of-concept and research agenda
Abstract
Emulation modeling (sometimes called "meta-modeling" or "surrogate modeling") refers to the practice of developing easier-to-use or less computationally expensive models that mimic the outputs of original simulation models. Increasingly sophisticated process-based hydrological simulation models have been able to capture important non-linear dynamics and spatial interactions between coupled surface, groundwater, and vegetative processes in hydrologically sensitive areas. However, despite their usefulness in scientific research, the direct application of complex, process-based models in land use planning and policy processes remains limited. For decision-making processes that occur in iterative or collaborative scenario-testing settings, two major barriers to the use of process-based models include: (1) their long simulation times and (2) high computational requirements. In this research, I demonstrate the use of deep learning algorithm-based models to emulate spatial and time-series outputs from the process-based hydrological model ParFlow.CLM. Designed for use within a Jupyter-notebook graphical user interface, the trained emulation model simplifies user interactions and has near-instantaneous simulation times, which can reduce key barriers to participatory stakeholder modeling. However, it may also worsen some aspects of the model's capacity to bridge boundaries between science and policy. To address these potential trade-offs, I propose a research agenda to evaluate the salience, credibility, and legitimacy of the use of emulation-modeling in participatory, spatial land use decision-making processes.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.H21O1963L
- Keywords:
-
- 1880 Water management;
- HYDROLOGY;
- 6344 System operation and management;
- POLICY SCIENCES;
- 6309 Decision making under uncertainty;
- POLICY SCIENCES & PUBLIC ISSUES;
- 6620 Science policy;
- PUBLIC ISSUES