AN OBJECTIVE ANALYSIS SCHEME FOR AVHRR IMAGERY

An objective analysis scheme has been modified and used for AVHRR imagery. The computing time involved in the sorting process applied to each interpolation point, has been significantly reduced exploiting the fact that satellite-derived data are regularly sampled and that the covariance function behaviour (monotonically decreasing) is known. Spatial gaps in the sea surface temperature (SST) maps resulting from cloud coverage, atmospheric phenomena and missing data were optimally interpolated using neighbouring, future and past pixels. The interpolated SST fields reproduce the relevant oceanographic features (and their evolution) but filter out the noise introduced by the split window atmospheric correction. Space and time correlation scales of 40 km and 10 days respectively were empirically determined as best choice for the correlation function parameters in the Ligurian Sea. These values are in agreement with previous in situ observations of meander wavelength and phase speed for the same area. Moreover, a comparison of objective analysis (OA) and compositing techniques results has shown that while the latter does not retain the rapidly varing features moving along the northern edge of the Ligurian gyre the former reproduces them accurately.