The quiet revolution of numerical weather prediction
- Bauer, Peter
- Thorpe, Alan
- Brunet, Gilbert
1 European Centre for Medium-Range Weather Forecasts, Shinfield Park, Reading RG2 9AX, UK
2 Environment Canada, Trans-Canada Highway Dorval, Québec H9P 1J3, Canada
c1Email:[email protected]
Received 19 February 2015; accepted 16 July 2015; Published online 2 Spetember 2015
Author Contributions P.B. led the writing of the paper and conceived the ideas of Figs 5 and 6. All authors contributed to writing the manuscript and assembled the remaining figures.
Figure 5 | CPU and power requirements as a function of NWP model resolution.
Simplified illustration of the number of compute cores (left y-axis) and power (in units of megavolt amps, MVA, right y axis) required for single 10-day model forecast (lower curves) and 50-member ensemble forecast (upper curves) as a function of model resolution, given today’s model code and compute technology. The shaded area indicates the range covered when assuming perfect scaling (bottom curve) and inefficient scaling (top curve), respectively. Today’s single global forecasts operate at around 15 km while ensembles have around 30 km resolution.
Figure 6 | Key challenge areas for NWP in the future.
Advances in forecast skill will come from scientific and technological innovation in computing, the representation of physical processes in parameterizations, coupling of Earth-system components, the use of observations with advanced data assimilation algorithms, and the consistent description of uncertainties through ensemble methods and how they interact across scales. The ellipses show key phenomena relevant for NWP as a function of scales between 10−2 and 104 km resolved in numerical models and the modelled complexity of processes characterizing the small-scale flow up to the fully coupled Earth system. The boxes represent scale-complexity regions where the most significant challenges for future predictive skill improvement exist. The arrow highlights the importance of error propagation across resolution range and Earth-system components.
Acknowledgements We thank C. Jakob and M. Miller for motivating us to write this paper. We are grateful to E. Källén, F. Rabier and A. Simmons for comments and to L. Magnusson for input and figures.
Author Information Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests
Advances in numerical weather prediction represent a quiet revolution because they have resulted from a steady accumulation of scientific knowledge and technological advances over many years that, with only a few exceptions, have not been associated with the aura of fundamental physics breakthroughs. Nonetheless, the impact of numerical weather prediction is among the greatest of any area of physical science. As a computational problem, global weather prediction is comparable to the simulation of the human brain and of the evolution of the early Universe, and it is performed every day at major operational centres across the world.