Weather Forecasting
Weather forecasting turns the laws of atmospheric physics and a flood of observations into predictions of tomorrow's sky, combining numerical models, data assimilation, and an honest accounting of uncertainty.
Definition
Weather forecasting is the application of meteorological knowledge and computational methods to estimate the future state of the atmosphere over time horizons from minutes to weeks, expressed deterministically or probabilistically.
Scope
This area covers the methods used to predict future weather, including the numerical models that integrate the governing equations forward in time, the data assimilation that anchors them to observations, the ensemble techniques that quantify predictability and uncertainty, and the statistical and nowcasting approaches used at short range.
Sub-topics
Core questions
- How are the governing equations of the atmosphere solved to predict its future state?
- How are scattered observations combined with models to define the starting conditions?
- Why is weather predictability limited, and how is forecast uncertainty quantified?
- How are short-range and statistical forecasts produced and verified?
Key theories
- Numerical weather prediction
- Weather can be forecast by integrating the discretized primitive equations forward from an observed initial state, an idea proposed by Bjerknes and Richardson and realized once electronic computers and data networks made it practical.
- Predictability and sensitive dependence
- Because the atmosphere is a chaotic system, small errors in the initial state grow over time, imposing an inherent limit on deterministic forecasting and motivating probabilistic, ensemble-based prediction.
Mechanisms
A forecast begins by estimating the current state of the atmosphere through data assimilation, which blends a short model forecast with new observations. A numerical model then integrates the governing equations forward in time on a grid, representing unresolved processes such as convection and radiation through parameterizations. Because initial errors grow, many forecasts are run from slightly different starting points and model configurations, forming an ensemble whose spread quantifies uncertainty; statistical post-processing and rapidly updated nowcasts refine predictions at short range.
Clinical relevance
Weather forecasts protect life and property and underpin aviation, agriculture, energy, water management, and disaster preparedness; the steady improvement of numerical prediction, often called a quiet revolution, has extended skillful forecasts by roughly a day per decade and made probabilistic guidance central to decision-making.
History
Vilhelm Bjerknes posed weather prediction as a calculable initial-value problem around 1904, and Lewis Fry Richardson attempted a hand calculation in the 1920s; the first successful numerical forecasts came with Charney and von Neumann's ENIAC computations around 1950. Lorenz's discovery of chaos in the 1960s reframed forecasting as inherently probabilistic, and growing computer power, global observations, and data assimilation produced today's ensemble-based operational systems.
Key figures
- Vilhelm Bjerknes
- Lewis Fry Richardson
- Jule Charney
- Edward Lorenz
Related topics
Seminal works
- kalnay2003
- lorenz1963
Frequently asked questions
- Why are weather forecasts less reliable beyond about a week?
- The atmosphere is chaotic, so tiny uncertainties in the starting conditions grow rapidly; after roughly one to two weeks these errors swamp the forecast, setting a practical limit on day-to-day predictability that no amount of computing power can remove.
- What does a percent chance of rain actually mean?
- It is a probabilistic forecast, often derived from an ensemble of model runs or statistical methods, expressing how likely measurable precipitation is at a location; a 30 percent chance means that, in similar situations, rain occurs about three times in ten.