Atmospheric Satellite Observations

Chapter 4

Filtering

Abstract

Observations always contain not only noise but also signals that could be true but are not resolvable at the analysis resolution. Interpolation that completes the task of mapping often irregularly distributed real-world data onto a model grid could generate computational and model errors (e.g., deviations from the dynamic constraints employed in the analysis). Although small scales in nature, these errors could induce significant large-scale errors in the analysis through aliasing or in numerical weather forecasts through computational instability. It is thus important to apply appropriate filters to remove errors arising from different sources when producing an analysis. Filtering is another required task of atmospheric ...