Premium
Variance spectra of observed winds and initialized wind fields at the 500 mblevel over europe
Author(s) -
Brown Philip S.,
Robinson G. D.
Publication year - 1985
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.49711146814
Subject(s) - initialization , variance (accounting) , scale (ratio) , national weather service , interpolation (computer graphics) , covariance , environmental science , wavenumber , range (aeronautics) , climatology , meteorology , mathematics , atmospheric sciences , statistics , geology , physics , geography , computer science , cartography , motion (physics) , materials science , accounting , classical mechanics , optics , business , composite material , programming language
Two‐dimensional scale (variance spectrum) analyses were made of the wind field at 500 mb over Europe and western Asia. Seven separate 20‐ to 30‐day periods, four in winter and three in summer, were investigated. On all occasions actual unmodified measured winds were used; on two of them the same techniques were applied to the interpolated and initialized data on a regular grid used by the National Weather Service in prognostic models. Standard analytical spectral analysis cannot be used on the unmodified observations; the method used involved iterative adjustment of a first guess spectrum to fit calculated variance and covariance functions. Some conclusions are: Our examination of the data does not unambiguously establish an ‘inertial range’ in the spectrum. the Taylor transformation of frequency to wavenumber is not reliable at scales more than about 1000 km. There are significant diurnal changes; in the summer periods they are not confined to the small unresolved scales. The Weather Service initialization‐interpolation procedure quantitatively removes variance at less than twice the grid scale, but also affects the spectrum in the resolved scales to a degree which might be significant in prognostic models. On two occasions there was evidence of widespread and persistent variance on a scale of about 350 km.