Lab in American Geophysical Union Fall Meeting 2015 in San Francisco

Date: 2015-06-30

Saman Razavi convenes the following two sessions at AGU Fall Meeting in San Francisco December 14-18. Deadline for abstract submisssion in August 5th, 2015.

Session ID#: 9796,  Session Title: Multiscale Dependency and Uncertainty in Modeling of Surface and Subsurface Environments

Session Description: Effective identification and modeling of hydrologic and biogeochemical systems requires scaleappropriate representations of the underlying processes. One challenge is that such processes occur at a range of spatiotemporal scales, which may be inconsistent with the scales of  application/model and available data (observations). For example, macroscopic models sometimes fail to adequately describe small-scale processes. This session invites contributions that address issues of scale and multiscale modeling, including (1) identification and representation of dominant surface processes across a range of scales (e.g., hillslope to watershed systems), (2) hybrid methods for coupling models at different scales and/or space-time resolutions (e.g., pore and continuum scales) in different parts of computational domains; (3) upscaling and downscaling strategies and transferability of parameters, state variables, observations, etc. across scales, (4) quantification and reduction of uncertainties across processes and scales, through data-model integration. Both surface and subsurface applications in hydrology and/or biogeochemistry are welcome.

Primary Convener:  Saman Razavi, University of Saskatchewan, Saskatoon, SK, Canada.
Conveners:  Timothy D Scheibe1, Zhangshuan Hou1 and Hoshin Vijai Gupta2, (1)Pacific Northwest National Laboratory, Richland, WA, United States(2)University of Arizona, Tucson, AZ, United States.


Session ID#: 8347 Session Title: Metamodeling and Surrogate Modeling: Addressing Model Uncertainty and Support for Decision Making

Session Description: Recent progress in surrogate- and meta-modeling provides opportunities to gain insight from complicated mathematical models in a fraction of the time. Model emulation can help identify sources of model uncertainty and leverage the ever-increasing level of detail inherent in modern models on a timescale meeting the practical needs of decision makers. In some cases, real-time operation and management guided by numerical models becomes possible. Statistical learning, artificial intelligence, model simplification, and other runtime mitigation accomplish these goals. Another value to surrogate/meta-modeling is coupling multiple processes more efficiently than iteratively linking. For example, climate and socioeconomic models, or groundwater and surface water models can be linked capturing the insights of each through surrogates.  We encourage contributions exploring new techniques and management endpoints for model simplification, emulation, metamodeling, and surrogate modeling of environmental processes. Techniques that propagate uncertainty from raw data, through modeling errors, to decision-support predictions and forecasts, are particularly encouraged.

Primary Convener:  Michael N Fienen, USGS Wisconsin Water Science Center, Middleton, WI, United States.

Conveners:  Anthony John Jakeman, Australian National University, Integrated Catchment Assessment and Management Centre, Canberra, Australia, Andrea Castelletti, Politecnico di Milano, Milano, Italy and Saman Razavi, University of Saskatchewan, Saskatoon, SK, Canada.