Three-dimensional variational data assimilation
Abstract
Observations can be directly assimilated in a three-dimensional variational (3D-Var) system as long as there is a forward model to generate simulations from a given model state. Dynamic and mathematical smoothing constraints can be either directly built into 3D-Var systems or added as penalty terms while assimilating newly available observations during a data assimilation cycle. This chapter shows that the 3D-Var formulation given in a deterministic format is equivalent to a statistical estimate of the maximum likelihood in which the background fields, observations, and models are “optimally” combined. Discussed are the National Meteorological Center method and recursive filters ...
Get Atmospheric Satellite Observations now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
