Homepage forecasts and outlooks
U.S. local forecasts and alerts originate with the National Weather Service. National severe-weather outlooks originate with the Storm Prediction Center; excessive-rainfall outlooks originate with the Weather Prediction Center. Surface temperature analysis is a different product from a station observation or a future forecast.
Labels distinguish observations, forecasts, and analyses. Source updates can arrive at different times. A refreshed page does not mean every provider has issued new data, so check the timestamp attached to the individual product.
Model Lab starts with numerical data
Model Lab reads the published catalog, run metadata, and compressed numerical grids from Mammatus data services. It does not recolor a competitor’s finished forecast map. The model’s available forecast hours determine the sequence; absent hours are not invented to fill the timeline.
Mammatus renders complete WebP graphics on the server. The browser downloads and decodes those graphics for playback. Pressure or height contours and supported wind barbs use the corresponding source fields when available. Automatic high and low markers are derived from the model’s sea-level pressure field; their detection and positions are estimates, not an official surface analysis.
Interpolation, city values, and missing data
Continuous shading uses interpolation to improve readability. Categorical precipitation-type information uses the source categories. Interpolation changes presentation; it does not increase the model’s physical resolution. Numeric city labels use nearby native grid values, which can differ from a local station or a terrain feature smaller than the grid.
Missing numerical values and source transparency rules remain unshaded. Pressure-level fields can contain terrain masks where a level lies below the ground. Mammatus does not fill those holes with fabricated values. Regional boundary lines and city labels are geographic references rather than additional forecast detail.
Coverage, comparisons, and archives
The model picker uses published grid footprints and regional-domain information to select models for the chosen region. Alaska has its own regional model option; a CONUS grid is not treated as Hawaii coverage. This is a coverage check, not a promise that every parameter contains a usable value at every point.
Comparisons align the products’ published valid times. Different run times therefore use different forecast leads. Archive discovery checks retained source products; the website cannot restore a run that the source has removed. Published source runs and locally cached images are different things.
Graphics, credits, and reuse
Forecast graphics retain initialization time, valid time, parameter, legend, and source attribution. Provider-specific copyright and licensing conditions are taken from the catalog. Mammatus processing does not imply provider endorsement. Keep those credits and timestamps when sharing a graphic, and consult the provider’s terms before a use beyond the displayed license.
Static exports at the current viewing size reuse the loaded graphic. High-resolution images and animations can require server rendering or encoding. The local cache is temporary and has size and age limits; it is not a permanent model archive.
Documentation and corrections
Our guides explain the website’s implemented behavior and use labeled examples. Source definitions are linked where they matter. Interactive teaching calculators run in your browser and make no weather prediction. They cannot establish whether an unpublished model field exists.
Send source or rendering concerns to [email protected] with the graphic’s model, run, parameter, forecast hour, region, and page address. See About Mammatus for our publishing approach, and the guides for worked examples.