Hierarchical physical-statistical forecasting in the atmospheric sciences

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] A class of hierarchical Bayesian models is introduced for Physical-Statistical forecasting purposes in the Atmospheric Sciences. The first project describes a methodological approach to implement a stochastic trigger function for convective initiation in the Kain-Fritsch (KF) convective parameterization scheme within the Penn State/NCAR Mesoscale Model version 5 (MM5). The second project introduces a spatio-temporal dynamic model that has a physical basis and incorporates Bayesian parameterizations and sequential importance sampling estimation to track and forecast the movement of multiple storm cells. The third project describes a finite difference model, in the framework of Bayesian hierarchical modeling (BHM), for investigating the possibility of forecasting the change of relative vorticity on a constant pressure surface in the middle troposphere over the globe.