Abstract
Spatial correlations for sea level pressures, during the Canadian Atlantic Storms Program of January-April 1986, were computed. We initially modeled these correlations using isotropic negative squared exponential, and second- and third-order autoregressive correlation functions. Allowances for variations due to latitude were made by partitioning the pressure stations north of 44°N from those south of 44°N and parameterizing observed correlations in each group using isotropic correlation functions. Tomporal variations were accommodated by separating pressure reports of the period early January to middle February from the period middle February to early April and parameterizing as before. Anisotropy was displayed by modeling the whole dataset using anisotropic second- and third-order autoregressive correlation functions.
These model correlation functions, which are used in optimal statistical estimation methods for the assimilation of data, were related to dynamical processes of the planetary boundary layer.
