Abstract The object of assimilation for NWP is to find an internally consistent model state which best fits all available information. This comes from remote sensing and in situ observations for the present time and the past few days; the NWP model is essential for the efficient use of earlier data. The determination of “best fit” must take account of the error characteristics of observations and model; Gaussian errors and a linearizable model give idealized equation with convenient properties, but gross errors and biases in observations are important, and spoil these. The growing diversity and volume of satellite data makes some pre-processing, before incorporation in NWP assimilations, essential. However it is important that the processing and assimilation is conceived as a single integrated process. The paper gives a brief summary of the theory, and approximate implementation, of the ideal equations, and how non-Gaussian errors affect this. Examples are considered from satellite temperature soundings, wind lidar, and radar scatterometer winds.
Assimilation of satellite data for global numerical weather prediction
Advances in Space Research ; 12 , 7 ; 285-294
01.01.1992
10 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
British Library Conference Proceedings | 1997
|Numerical Solution of the Variational Data Assimilation Problem Using Satellite Data
British Library Conference Proceedings | 2010
|