NOAA HRRR, 3 km. Every model grid cell is drawn as its own shape at its exact position (corners from the model's Lambert conformal projection), so cells look like cells at any zoom: zooming never adds model detail. "Smooth display" blends colours between cell centres for readability only; it never crosses missing data, and point values always come from the original cells. Wind speed is calculated from compatible 10-m wind components; gusts are the separate surface gust diagnostic at the valid hour. Total cloud cover is for the entire atmosphere, not cloud ceiling. Composite reflectivity is the model's column maximum; 1 km reflectivity is the model at 1,000 m above ground; both are simulated, not radar. Precipitation windows are read from the GRIB accumulation intervals and are blank unless every hour is present. The closest zoom is limited to a Cincinnati-area regional view. City labels are deliberately sparse; venue labels appear when selected. Roads are shown without route-number badges. Weather fields download only when selected; up to two runs are kept in memory for quick switching. Geography: US Census boundaries (us-atlas) and Natural Earth (public domain); towns from GeoNames (CC BY 4.0). Model guidance only, not an official forecast.