Numerical weather prediction Flashcards
What is the smallest horizontal grid-spacing currently used by the UK Met Office in an operational forecast model for the UK
This is currently a 1.5km grid-spacing in the short range UK forecast. This allows good resolution of our hills and coast lines, and allows moist convective storms to be explicitly captured by the grid.
how does the smallest horizontal grid used by UK compare to the global forecast model?
This compares with around 20km for the global forecast model, where such storms must be parametrised, and features such as fronts are much more crudely resolved. In many climate models the grid-spacing is around 100km, providing even less resolution of hills, coasts and fronts.
“The atmosphere is a chaotic system” What does this mean?
A chaotic system is one where the evolution of the system follows deterministic laws, but small differences in initial conditions can lead to large differences in the final result, limiting the predictability of the system. In this way a system that is not random is still unpredictable. For example, a small error in in initial conditions for a forecast can mean that a one-day forecast is quite accurate, but make a two-week forecast for the same situation very difficult, as the small error grows to produce a large uncertainty after that time.
- What is data assimilation in Numerical Weather Prediction?
Data assimilation is the process by which observations are ingested into a very short range weather forecast to determine our “best guess” of the current state of the atmosphere (‘the analysis’). The forecast model is then initialized with the analysis and run forward in time to give the weather forecast. There is still uncertainty in the analysis, as the observations and short range forecast both have uncertainties, and there are always many points in the atmosphere which are unobserved.
- Explain in numerical weather prediction (NWP) what ‘resolved’ and ‘parameterised’ mean.
In an NWP system the atmosphere is divided up using a grid. The flow is then calculated on this grid, so features that are many times the grid-spacing are ‘resolved’. Some processes occur on finer scales (e.g. transfer of solar and infrared radiation, cloud droplets combing to form rain, melting of falling snow etc) and these must be included in models by physically based approximations of these processes, i.e. ‘parameterised’.