Handling Model Uncertainty: The Importance of Human Knowledge
Abstract
Most approaches for addressing model uncertainty focus on finding new ways to better quantify the various components that contribute to uncertainty (measurement error, parameter error, model error, etc.). If this is the sole basis for creating model forecasts under uncertainty, the resulting error bounds can be so large as to make the forecasts useless. This paper will illustrate the importance of human knowledge in narrowing the number and types of potential models and decision strategies that are considered, focusing on case studies in groundwater model calibration and monitoring design. An interactive optimization framework is presented that balances expert judgment with traditional quantitative objectives such as model error. Results show how the expert's input shifts model and management strategy selection to more reasonable solutions that still perform extremely well on traditional error metrics. Ideas for extending the research to include multiple stakeholders and robust optimization will also be presented.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFM.H24D..02M
- Keywords:
-
- 1846 HYDROLOGY / Model calibration;
- 1848 HYDROLOGY / Monitoring networks;
- 1873 HYDROLOGY / Uncertainty assessment;
- 1880 HYDROLOGY / Water management