Assimilation of satellite data in numerical weather prediction. Part I: The early years

Developments in the assimilation of satellite data in numerical weather prediction (NWP), from the first experiments in the late 1960s to the present day, are presented in a two‐part review article. This first part reviews the early years, up to about the year 2000. It includes summaries of the relevant satellite remote sensing technologies, the theoretical and practical challenges faced when assimilating their data within NWP systems, and the impacts on forecast skill. An important part of this story concerns developments in the assimilation of information on atmospheric temperature and humidity provided by data from passive infrared and microwave radiometers. Following early successes with the assimilation of retrieved temperature profiles, there followed a problematic period, as other aspects of NWP systems improved and the impacts of satellite sounding data declined. Positive impacts were re‐established in the 1990s through moves towards more direct assimilation of radiance information. Another important theme concerns developments in the assimilation of wind information via atmospheric motion vectors, which underwent a series of improvements during these years. Additional contributions were provided by information on ocean surface wind from scatterometers. Some contributions from other technologies during this period are also summarised. TOVS scan patterns for HIRS (small dots) and MSU (ellipses), from Smith et al. (1979).