We have evaluated E from the NCEP reanalysis data (Kalnay et al. 1996) and T from the Hadley Centre monthly sea surface temperature data (Rayner et al. 2003) by averaging both sets of data over each month. The spatial resolution of the Hadley Centre data is 10, and the NCEP data, which are on a T62 Gaussian grid were interpolated on to the Hadley Centre grid. The gradient, dlnE/dT, was then computed at all sea points for which the first order finite-difference approximations of the spatial derivatives could be evaluated. Thus gaps in the field of H occur around islands and along the continental coastlines. 27 years (1979-2005) of data were available for analysis.
Results were obtained for three northern hemisphere regions in which tropical cyclones are known to be generated and one reference region in the southern subtropics (Table 2). The number of grid point—months in each region was approximately 20,000, giving a large sample for statistical analysis, from which the mean and standard deviation of H values were obtained by averaging over all grid points and all years for each month. Histograms of the distribution of H by 0.1 K—1 class interval, and the mean and standard deviation of H over the range, —3 < H < 3 K—1 were also computed. This was done because the evaluation of H from proposal next may contain 'outliers' of large magnitude due to the interpolation procedure, which were not representative of the physics.
Was this article helpful?