E-Book Overview
Lewis Fry Richardson dreamt that scientific weather prediction would one day become a practical reality. Before his ideas could bear fruit several advances were needed: better understanding of the dynamics of the atmosphere; stable computational algorithms to integrate the equations; regular observations of the free atmosphere; and powerful automatic computer equipment. By 1950 advances in all these fronts were sufficient to permit the first computer forecast to be made. Over the ensuing fifty years progress in numerical weather prediction has been dramatic. Weather prediction and climate modelling have now reached a high level of sophistication. This book tells the story of Richardson's trial forecast, and the fulfilment of his dream of practical numerical weather forecasting. It includes a complete reconstruction of Richardson's forecast, and analyses in detail the causes of his failure. This will appeal to everyone involved in numerical weather forecasting, from researchers and graduate students to professionals.
E-Book Content
Richardson’s Dream The Emergence of Scientific Weather Forecasting
Peter Lynch
Contents
Guiding Signs Preface Acknowledgements
page vi vii ix
1
Weather Prediction by Numerical Process 1.1 The problem 1.2 Vilhelm Bjerknes and scientific forecasting 1.3 Outline of Richardson’s life and work 1.4 The origin of Weather Prediction by Numerical Process 1.5 Outline of the contents of WPNP 1.6 Preview of remaining chapters
1 1 4 10 14 18 26
2
The fundamental equations 2.1 Richardson’s general circulation model 2.2 The basic equations 2.3 The vertical velocity equation 2.4 Temperature in the stratosphere 2.5 Pressure co-ordinates
28 29 30 38 41 43
3
The oscillations of the atmosphere 3.1 The Laplace tidal equations 3.2 Normal modes of the atmosphere 3.3 Atmospheric tides 3.4 Numerical solution of the Laplace tidal equations
46 47 48 54 55
4
The barotropic forecast 4.1 Richardson’s model and data 4.2 The finite difference scheme 4.3 The barotropic forecast 4.4 The global numerical model 4.5 Extending the forecast
62 62 64 67 70 71
iii
iv
Contents
4.6 4.7
Non-divergent and balanced initial conditions Reflections on the single layer model
74 75
5
The solution algorithm 5.1 The finite difference method 5.2 Integration in time 5.3 The Courant-Friedrichs-Lewy Stability Criterion 5.4 The Richardson grid 5.5 The equations for the strata 5.6 The computational algorithm
78 78 80 84 87 88 91
6
Observations and initial fields 6.1 Aerological observations 6.2 Dines’ meteorograph 6.3 The Leipzig charts 6.4 Preparation of the initial fields
95 95 98 102 108
7
Richardson’s forecast 7.1 What Richardson actually predicted: twenty numbers 7.2 Scaling the equations of motion 7.3 Analysis of the initial tendencies 7.4 The causes of the forecast failure 7.5 Max Margules and the ‘impossibility’ of forecasting
116 116 119 124 130 132
8
Balance and initialization 8.1 Balance in the atmosphere 8.2 The slow manifold 8.3 Techniques of initialization 8.4 The swinging spring 8.5 Digital filter initialization
135 135 138 141 144 150
9
Smoothing the forecast 9.1 Reconstruction of the forecast 9.2 Richardson’s five smoothing methods 9.3 Digital filtering of the initial data 9.4 Extension of the forecast
156 156 159 161 172
10
The ENIAC integrations 10.1 The ‘Meteorology Project’ 10.2 The filtered equations 10.3 The ENIAC integrations 10.4 The barotropic model 10.5 Multi-level models 10.6 Primitive equation models
178 179 184 187 193 196 199
Contents
v
10.7 General ci