ACKNOWLEDGEMENTS

I’d like to express my gratitude to many of my colleagues at the Met Office who I have pestered for information and explanations, in particular David Richardson, Richard Dumelow and Colin Parrett. I am also appreciative for the critical comments and guidance from my dissertation supervisor, Alan Thorpe. Further I would like to thank the ‘Met Office ’ for allowing me time to complete this work, and also for turning a blind eye to the chaotic state of my desktop! And finally I’d like express my gratitude to my wife, Everlyn, for her support and patience during this work. i An accurate analysis of the atmospheric state is a vital ingredient for skilful numerical weather predictions. This dissertation examines the particular role of weather observations in defining the initial atmospheric conditions for global numerical weather predictions and attempts to deduce their future requirements. Current observational requirements for global numerical weather prediction are met, to varying degrees, by a range of terrestrial and space-based observing systems. The characteristics of thes