Variance Characteristics of Tropical Radiosonde Winds Using a Vector-Tensor Method

Abstract

In this paper, an exploratory study on the variance characteristics of upper-air winds in a near-equator monsoon region is presented. The data were obtained from historical radiosonde observations from up to 250 stations within the region of interest for the period between 1954 and 2013. An alternative method based on vector statistics was employed in this study which characterises the mean by a vector and the variance by a tensor. Unlike the conventional approach of using scalar wind speeds, this vector-tensor approach allows the directional properties of the variance to be studied. A suite of statistics to describe the geometric properties of the variance tensor was also developed. These characterise the size of the variance, its degree of anisotropy, and the alignment of the preferred direction (if anisotropy is present) with the direction of the mean wind vector. Through analysis of these statistics, several salient trends were observed for the middle troposphere. It was found that the variance size and anisotropy exhibit significant variation with height whereas the alignment with the mean vector varies with the mean wind magnitude instead. It was also found that the scalar variance increases with mean wind speed.

Keywords

• wind characteristics
• variance
• covariance matrix
• random vector
• radiosonde
• monsoon
• tropical
• high-altitude