Statistics for Geophysical Models

Douglas W. Nychka
National Center for Atmospheric Research

A touchstone for the geosciences is the understanding and prediction of physical systems through numerical models. Using modern statistical methods it is possible to combine numerical models with statistical analysis to produce a synergy between physical and data-based modeling. For numerical models this adds more flexibility in representing complex subgrid scale phenomenon as stochastic processes. For statistical research there is the challenge of finding compact and useful statistical descriptions based on numerical model output. This talk gives examples of the role of statistics across a range of temporal and spatial scales of geophysical processes: developing parameterizations and analyzing output of global climate models, combining data and model output for weather forecasts and using model output to predict clear air turbulence.

Return to the Program