Abstract
On the basis of a 10-yr record of rawinsonde observations in the Tropics, experiments were run to illustrate the manner in which climatology may be used to minimize the root-mean-square errors of interpolation from data of mixed quality that are irregularly located in time and space. The procedure, based on the theory of optimum interpolation, determines the relative weights of the data used in the interpolation on the basis of their error characteristics, their location, and the scale and variability of the meteorological fields that they sample.
