Functional representation of observed correlations for statistical interpolation scheme of objective analysis

Abstract

For statistical interpolation scheme of objective analysis, the weights of matrices are determined from the correlation functions of meteorological parameters, which are fitted to a function to facilitate computations. In this regard, the horizontal height correlations of observed height residuals (differences between radiosonde data and the 24 hrs forecast that are used as initial fields) are represented by a sum of two degenerated third-order auto-regressive (TOAR) functions after suitably modifying it for the Indian region. The geostrophic approximation, relating height and wind field is decoupled near the equator. Analyses were made for two different synoptic situations, viz. 4-8 July 1979 and 26-30 July 1991 for 1200 hrs GMT. It was found that the analyses are closer to the actual situations, as the root mean square errors for new scheme were less in both the situations as compared to the analyses made with the Gaussian correlation function (old scheme), representing the height-height correlations.