A climatology of tropical wind shear produced by clustering wind profiles from the Met Office Unified Model (GA7.0)

Abstract

Toward the goal of linking wind shear with the mesoscale organization of deep convection, a procedure for producing a climatology of tropical wind shear from the output of the Met Office Unified Model climate model is presented. Statistical information from wind profiles from tropical grid columns is used to produce a tractable number (10) of profiles that efficiently span the space of all wind profiles. Physical arguments are used to filter wind profiles that are likely to be associated with organized convection: only grid columns with substantial convective available potential energy (CAPE) and those with shear in the upper quartile are considered. The profiles are rotated so that their wind vectors at 850 hPa are aligned, in order to be able to group like profiles together, and their magnitudes at each level are normalized. To emphasize the effect of lower levels, where the organization effects of shear are thought to be strongest, the profiles above 500 hPa are multiplied by 14. Principal component analysis is used to truncate the number of dimensions of the profiles to seven (which explains 90 % of the variance), and the truncated profiles are clustered using a K-means clustering algorithm. The median of each cluster defines a representative wind profile (RWP). Each cluster contains information from thousands of wind profiles with different locations, times and 850 hPa wind directions. To summarize the clusters statistically, we interpret the RWPs as pseudo-wind profiles and display the geographic frequency, seasonal frequency and histograms of wind direction at 850 hPa for each cluster. Geographic patterns are evident, and certain features of the spatio-temporal distributions are matched to observed distributions of convective organization. The form of the RWPs is also matched to specific wind profiles from case studies of organized convection. By performing the analysis on climate-model output, we lay the foundations for the development of the representation of shear-induced organization in a convection parametrization scheme (CPS). This would use the same methodology to diagnose where the organization of convection occurs and modify the CPS in an appropriate manner to represent it. The procedure could also be used as a diagnostic tool for evaluating and comparing climate models.