A univariate statistical scheme for the analysis of geopotential height fields has been developed based upon optimum interpolation. The data archived from a set of subjective analyses were used to compute the required correlation functions. The computed height autocorrelation functions were fitted by an appropriate function which would definitely reflect the characteristics of the statistical structure in the low latitudinal regions. Rather than representing the model by negative squared exponential form proposed by Gandin (1963), representation consisting of a polynomial exponential function was used to fully trace the finer structure of the statistical quantities. The analyzed geopotential height field were obtained through the computation of the geopotential height autocorrelation model. As a measurement of accuracy, this scheme has been tested at target stations. The root mean squares between the analysed data and those obtained form radiosonde data have been carried out toa verify its performance as well.