Premium
A method for multiscale optimal analysis with application to A rgo data
Author(s) -
Gray Alison R.,
Riser Stephen C.
Publication year - 2015
Publication title -
journal of geophysical research: oceans
Language(s) - English
Resource type - Journals
eISSN - 2169-9291
pISSN - 2169-9275
DOI - 10.1002/2014jc010208
Subject(s) - argo , a priori and a posteriori , scale (ratio) , algorithm , computer science , least squares function approximation , mathematics , mathematical optimization , data mining , statistics , physics , geology , philosophy , oceanography , epistemology , quantum mechanics , estimator
This study presents an optimal analysis method for estimating the large‐ and small‐scale components of a field from observations. This technique relies on an iterative generalized least squares procedure to determine the statistics of the small‐scale fluctuations directly from the data and is thus especially valuable when such information is not known a priori. The use of spherical radial basis functions in fitting the large‐scale signal is suggested, particularly when the domain is sufficiently large. Two test cases illustrate several of the properties of this procedure, demonstrate its utility, and provide practical guidelines for its use. This method is then applied to observations collected by the Argo array of profiling floats to produce global gridded absolute geostrophic velocity estimates.