Experiments with statistical objective analysis techniques for representing a coastal surface temperature field

For interpretation of thermally driven mesoscale circulation systems, meteorologists are interested in detailed structures of boundary-layer temperature fields. The surface temperature of a coastal region is analyzed using an optimum interpolation scheme. Covariance estimates are frequently based on the assumption of isotropy. The interpolation is improved by consideration of the anisotropic character induced by inhomogenities of the terrain. By a simple model, typical variations of the covariance function in a coastal zone are estimated. The results are discussed by a comparison of interpolated and observed time series. Finally the refined statistical model is tested by the application to objective analysis of coastal surface temperature fields. It is evident that the specification of topographic dependencies leads to a reduction of mean square interpolation error.