Climate Model Calibration Across Multiple Spatial Resolutions

• Activity Number: 264
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Vera Bulaevskaya* and Donald Lucas
• Companies: Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory
• Keywords: climate modeling ; model calibration ; computer experiments ; multi-fidelity computer simulators

Abstract

Parameterizations of physical processes in climate models are highly dependent on the spatial and temporal resolution and must be tuned for each resolution under consideration. At high spatial resolutions, objective methods for parameter tuning are computationally prohibitive. Our work has focused on calibrating parameters in the Community Atmosphere Model 5 (CAM5) at three spatial resolutions: 4, 2 and 1 degrees. Using perturbed-parameter ensembles and uncertainty quantification methodology, we have identified input parameters that minimize discrepancies of key output quantities simulated by CAM5 across these three resolutions and with respect to satellite data. We are also beginning to exploit the parameter-resolution relationships to calibrate the high-resolution version of CAM5 by leveraging cheaper, low-resolution simulations and statistical models.