Bayesian Modeling in the Wavelet Domain with Applications in Atmospheric Turbulence

点击翻译按钮获取中文摘要

This proposal builds on a previous NSF sponsored project DMS-9626159, awarded to Duke University. It involves development of statistical concepts, theories and methods for multiresolution models and analyses of measurements and structures in atmospheric turbulence. It draws on recent developments in Bayesian multiscale modeling to understand the time-scale aspects of turbulence where the notions of scale and hierarchy are intrinsic. Multiresolution statistical modeling approaches are applied to vast geophysical measurements arising from air quality field experiments and simulations in order to identify key structural properties of atmospheric turbulence responsible for the transport of scalars, such as ozone. To achieve these objectives a team of researchers with expertise in statistical modeling and multiscale methods (M. Pensky and B. Vidakovic) and measurement and modeling of atmospheric turbulence (G. Katul) is assembled.The aplicability of this project is improved fundamental understanding of atmospheric transport via novel statistical techniques. Atmospheric transport, key to describing air-quality, is a complex phenomena involving modeling of turbulence which is responsible for the dispersion of polutants. The applicability of the proposed methodology will be directly tested on high frequency velocity, temperature, and ozone concentration measurements collected at Duke Forest, Durham, North Carolina. Partial support for this award is provided by the Physical Meterology Program in the Division of Atmospheric Sciences.