NOAA GFS

The NOAA Global Forecast System (GFS) provides global weather forecast data at 0.25° (~25 km) resolution. Data is accessed via the NOMADS grib filter, which allows subsetting GRIB2 files by region, variables, and levels — returning smaller GRIB2 files over HTTP.

No API key is required. A ~10 second delay between requests is recommended.

MetaData(NOAAGFS())

MetaData("", "~10 sec between requests", :weather, Dict(:ABSV => "Absolute vorticity (1/s)", :APCP => "Total precipitation (kg/m²)", :PRATE => "Precipitation rate (kg/m²/s)", :CIN => "Convective Inhibition (J/kg)", :PWAT => "Precipitable water (kg/m²)", :VGRD => "V-component of wind (m/s)", :TCDC => "Total cloud cover (%)", :RH => "Relative humidity (%)", :PRMSL => "Pressure reduced to MSL (Pa)", :TMP => "Temperature (K)"…), :raster, "0.25° (~25 km)", "Global", :forecast, nothing, "Forecast (4 cycles/day, 0–240 h)", "Public Domain", "https://nomads.ncep.noaa.gov/", Dict("Rasters" => "a3a2b9e3-a471-40c9-b274-f788e487c689"))

Basic Usage

Unlike most sources, GFS is forecast data — you specify a model run_date and cycle instead of a date range, plus forecast_hours for which forecast steps to retrieve.

using GeoInterface.Extents: Extent
using Dates

# Temperature and wind for the central US, today's 00z run, analysis + 3h + 6h
plan = DataAccessPlan(NOAAGFS(),
    Extent(X=(-100.0, -90.0), Y=(35.0, 45.0));
    variables = [:TMP, :UGRD, :VGRD],
    levels = ["2_m_above_ground"],
    forecast_hours = [0, 3, 6])
files = fetch(plan)

Each forecast hour produces a separate GRIB2 file.

Specific Model Run

plan = DataAccessPlan(NOAAGFS(),
    Extent(X=(-120.0, -80.0), Y=(30.0, 50.0));
    run_date = Date(2025, 6, 15),
    cycle = 12,
    variables = [:TMP, :RH, :PRATE],
    levels = ["surface", "2_m_above_ground"],
    forecast_hours = [0, 6, 12, 24])

Options

Keyword	Type	Default	Description
`run_date`	`Date`	`today()`	Model run date
`cycle`	`Int`	`0`	Model cycle hour (0, 6, 12, or 18)
`forecast_hours`	`Vector{Int}`	`[0]`	Forecast steps to retrieve (0–240)
`variables`	`Vector{Symbol}`	`[:TMP, :UGRD, :VGRD]`	GFS variable names
`levels`	`Vector{String}`	`["2_m_above_ground", "10_m_above_ground"]`	Vertical levels

Common Variables

Variable	Description
`TMP`	Temperature (K)
`UGRD`	U-component of wind (m/s)
`VGRD`	V-component of wind (m/s)
`RH`	Relative humidity (%)
`PRATE`	Precipitation rate (kg/m²/s)
`APCP`	Total precipitation (kg/m²)
`PRMSL`	Pressure reduced to MSL (Pa)
`DSWRF`	Downward short-wave radiation flux (W/m²)
`DLWRF`	Downward long-wave radiation flux (W/m²)
`CAPE`	Convective Available Potential Energy (J/kg)
`TCDC`	Total cloud cover (%)
`PWAT`	Precipitable water (kg/m²)
`HGT`	Geopotential height (gpm)

Common Levels

Level	Description
`surface`	Ground surface
`2_m_above_ground`	2 m above ground (standard temperature)
`10_m_above_ground`	10 m above ground (standard wind)
`entire_atmosphere`	Entire atmospheric column
`mean_sea_level`	Mean sea level
`500_mb`, `850_mb`, etc.	Pressure levels (1000–10 mb)

Loading GRIB2 Data

fetch returns file paths to GRIB2 files. To load them as raster data, install the Rasters package:

using Pkg
Pkg.add("Rasters")

using Rasters
files = fetch(plan)
r = Raster(files[1])