Intercomparison of Methods for the Temporal Interpolation of Synoptic Wind Fields

Abstract

The problem of temporal interpolation of wind fields is addressed by comparing the performance of standard linear interpolation with two methods that aim to provide a more accurate description of advecting weather systems: the complex empirical orthogonal function (EOF) method introduced by Zavala-Hidalgo and a method based on fast Fourier transform (FFT) techniques. Two test cases are considered. In the first, wind fields representing an idealized fast-moving tropical cyclone were interpolated. In this case, the FFT method provided root-mean-square errors approximately two-thirds of those from linear interpolation, while the EOF method produced larger errors than linear interpolation. The second test case, using one month of ECMWF analysis fields on a Southern Hemisphere regional domain, showed that the FFT method is insufficiently robust for complex wind fields with multiple moving weather systems. The EOF method, on the other hand, produced smaller errors than linear interpolation.