Home/Field guides/How a model advances the atmosphere
Forecasts & uncertainty / FIELD GUIDE

How a model advances the atmosphere

A forecast begins with an estimated atmospheric state and evolves it using a numerical representation. Every stage introduces limits.

NowLaterThreshold
Illustration of a related weather process.

What to notice

Two forecast maps agree on a broad low but place its narrow rain band differently.

What is happening?

Numerical weather prediction represents the atmosphere on a grid or another computational structure and advances its state through time. Equations describe motion, heat, moisture and related processes. Observations from stations, balloons, satellites and other platforms help construct the initial analysis. That analysis combines imperfect measurements with a previous model estimate; it is not a complete direct observation of every point in the atmosphere.

Processes smaller than the resolved scale must be represented approximately. Clouds, turbulence and surface exchange interact with the larger flow, so those approximations influence the forecast. A finer grid can represent more detail, but it does not automatically remove errors in initial conditions or physical assumptions. A very detailed-looking map can still place a feature incorrectly.

Basis: NOAA / National Weather Service [1] · ECMWF [2] · ECMWF [3]

Read the observation

  • Read the model run time separately from the forecast valid time.
  • Check resolution and whether the field was statistically adjusted.
  • Compare broad patterns before focusing on individual grid cells.
A USEFUL DISTINCTION

Higher map detail is not proof of higher accuracy, and a model analysis is not the same thing as a map of raw observations.

Go a little deeper

Forecast uncertainty arises from incomplete knowledge of the initial atmosphere and from limitations in the model. Small differences can grow differently in different weather situations. There is no single universal day after which every forecast becomes useless. Predictability depends on the variable, spatial scale, season and evolving flow; a broad temperature tendency can remain informative when a specific shower location does not.

For a controlled experiment, start two simplified scenarios with slightly different assumptions and compare the outcomes at the same future time. The probability lab does this transparently without claiming to solve the full atmospheric equations. In real interpretation, use ensembles and official forecast context to assess whether a single model image represents a robust signal or one of several plausible scenarios.

Basis: NOAA / National Weather Service [1] · ECMWF [2] · ECMWF [3]

FOLLOW THE EVIDENCE

Sources & provenance

  1. Weather Models NOAA / National Weather Service · Publication date not statedReference only; original paraphrase; no source media reproduced. JetStream permits broad use of its materials; separate credits are not assumed cleared.
  2. Quantifying forecast uncertainty ECMWF · Publication date not statedReference only; original paraphrase; no source media reproduced.
  3. Fact sheet: Ensemble weather forecasting ECMWF · Publication date not statedReference only; original paraphrase; no source media reproduced.

Explanations and diagrams are original educational interpretations. Source institutions have not endorsed or reviewed this publication. How we distinguish evidence from models.