Oceanographic data assimilation and regression analysis

✍ Scribed by Keith R. Thompson; Michael Dowd; Youyu Lu; Bruce Smith

• Publisher: John Wiley and Sons
• Year: 2000
• Tongue: English
• Weight: 158 KB
• Volume: 11
• Category: Article
• ISSN: 1180-4009
• DOI: 10.1002/(sici)1099-095x(200003/04)11:2<183::aid-env401>3.0.co;2-h

✦ Synopsis

A simple method is described for assimilating a set of irregularly spaced observations into a dynamicallybased model of the coastal ocean. The method can be used with complex models of high dimension and is relatively ecient and eective. It is based on the use of a simpler model to reduce, in an iterative fashion, the mean square dierence between the observations and the predictions of the complex model. To illustrate the method we use it to predict tidal sea-levels and currents in the Gulf of St. Lawrence, a semienclosed sea o Canada's east coast, from sea-levels measured by 19 coastal tide gauges. The method is shown to predict sea-levels to within several cm, and currents to within several cm s À1 . To explain the method, we relate it to the familiar concept of nonlinear regression and the Gauss±Newton algorithm for the minimization of a multivariate function.