Empirical Orthogonal Function Estimates of Local Predictability

Open Access
Publisher Website
Google Scholar
Abstract
Predictability times are estimated from 500 mb heights resolved for horizontal scales from 5 × 106 to 8 × 106 m. These scales are defined through eigenvector analysis of seven years of winter data over a portion of the Northern Hemisphere centered over the western United States. An analysis of variance suggests characteristic times (T0) of 6–12 days for effectively independent heights. A random process is chosen which fits well the frequency spectra of the time-dependent coefficients. The estimates of T0 calculated from this random model are about half of those obtained from the variance analyses. These differences might be due to sampling fluctuations as to the existence of long-period oscillations poorly represented by the random process. Both methods show a decrease in T0 with decreasing horizontal scale. Abstract Predictability times are estimated from 500 mb heights resolved for horizontal scales from 5 × 106 to 8 × 106 m. These scales are defined through eigenvector analysis of seven years of winter data over a portion of the Northern Hemisphere centered over the western United States. An analysis of variance suggests characteristic times (T0) of 6–12 days for effectively independent heights. A random process is chosen which fits well the frequency spectra of the time-dependent coefficients. The estimates of T0 calculated from this random model are about half of those obtained from the variance analyses. These differences might be due to sampling fluctuations as to the existence of long-period oscillations poorly represented by the random process. Both methods show a decrease in T0 with decreasing horizontal scale.

This publication has 0 references indexed in Scilit: