Empirical analysis of the non-Gaussianity of global geopotential heights

Since the beginning of modern atmospheric dynamics, it has been assumed that for large scales the distribution of atmospheric variables is roughly Gaussian or normal. This assumption resulted in the mathematical (and computational) convenience that, after linearizing the main equations of motion, we could get rid of the higher order terms. However, by the 1980s enough data was available to convincingly show that atmospheric flow is largely non-Gaussian (see White 1980, Trenberth and Mo 1985, Nakamura and Wallace 1991, and Holzer 1996.) It is therefore necessary to develop (1) a better physical understanding of the forces at work and (2) a more robust dynamical model that can predict extreme events in climate.

Though this research project aims to achieve both of these tasks, this poster will focus only on the first one. To better understand the non-Gaussianity of atmospheric flow, sixty years (1948 to 2007) of daily NCEP / NCAR Reanalysis geopotential height data are analyzed empirically. The spatial scope of the data (17 vertical levels going from the surface to the mid-stratosphere, each level containing 144 * 73 horizonal grid points) allows us to substantially improve on previous knowledge by building a coherent three-dimensional picture of geopotential height skewness and kurtosis. The tropospheric pattern features negative skewness at mid-latitudes and positive skewness at high latitudes that are mostly due to the motion of planetary waves. Amplifications in the skewness pattern are due to atmospheric blocking and tropical cyclones. It is found that while inter-seasonal and inter-hemispherical variability in higher statistical moments is low in the troposphere, it is substantially higher in the stratosphere. Further analysis shows that this variability is due to stratospheric sudden warming events, which are caused by vertically propagating Rossby waves in the winter season. This data supports the hypothesis that interactions between the troposphere and stratosphere are influential in large-scale extreme events.