The detection of hurricane asymmetries from aircraft reconnaissance flight data: Some simulation experiments

Abstract

AbstractA series of observational system simulation experiments designed to investigate the data requirements for determining the flow asymmetry of a hurricane are described. Of particular interest is the azimuthal wave‐number‐one component of the asymmetry, which, together with the mean flow across the storm, is responsible for the storm motion. A principal finding is that the accuracy of retrieval of this component from wind data obtained from airborne omega dropwindsonde soundings depends on the orientation of aircraft flight patterns relative to that of the asymmetry. Since the orientation of the asymmetry is not known in advance, there will be an inherent uncertainty in the accuracy with which the asymmetry is determined. The simulation experiments enable one to estimate probable errors in the determination of the asymmetry and their implications for hurricane track errors.The track errors are assessed by comparing the vortex track in a high‐resolution control experiment with those that result from initial analyses based on simulated data sets of varying data density and quality. In some of the experiments, an increase in resolution along the flight tracks, or additional data in the storm environment, does not lead to an improvement in the forecast track, in contrast to increased data coverage on a regular grid. This at first sight surprising result is likely to be a consequence of the constant influence radius used in the objective analysis scheme, a factor which is currently being investigated.The addition of random wind‐measurement errors to the simulated data, comparable in magnitude to those in dropwindsonde measurements, does not lead to significantly inferior analyses or track forecasts based on them, suggesting that such errors are not the major cause of inaccuracy in diagnosing the vortex asymmetry. Moreover, the accuracy of the analysis is nearly independent of whether the data are taken at an instant of time, or over a typical nine‐hour flight duration and then corrected to a storm‐centred coordinate system.