Times are displayed in (UTC-06:00) Mountain Time (US & Canada)Change
201-4 Ensemble Methods for Assimilating Subsidence Information: How the Sausage is Made
Session: Land Surface Subsidence—Processes, Impacts, and Ongoing Challenges
Presenting Author:
Jeremy White
Author:
White, Jeremy1r>
(1) INTERA, Inc, ,
Abstract:
Land subsidence in Central Valley has been studied extensively and monitored at many sites for decades. Still, many efforts to simulate the historic and possible future patterns of land subsidence struggle to appropriately represent observed historic subsidence and compaction, a potential “red flag” for model predictive skill. These struggles are likely due to a combination of factors including: model scale and resolution, subsidence process representation and simplification, as well as inverse problem design and implementation.
This talk will introduce some concepts around model purpose, design, and simplification, then present some theory and concepts related to using ensemble methods to assimilate subsidence and compaction information, with care given to parameterization, objective function design, and coping with nonlinearity. A case study from a simple 1-D column model of a long-term monitoring site will be used to demonstrate concepts.