GeoFetch

Fetch spatio-temporal data in a specific region of interest from multiple sources.

Quick Start

using GeoFetch, Dates, Extents

gfs = GeoFetch.GFS_025  
gfs.parameters = ["TMP"]
gfs.levels = ["2_m_above_ground"]

p = Project(
    geometry = Extent(X=(-90.0, -80.0), Y=(30.0, 40.0)),
    datetimes = (DateTime(2026, 4, 1), DateTime(2026, 4, 1)),
    datasets = [gfs],
)

fetch(p)

Type Hierarchy

• Source -- a data provider (e.g. NOMADS, CDS, FIRMS)
  • datasets(source) -- list available datasets
  • help(source) -- documentation URL
• Dataset -- a configurable remote dataset
  • chunks(project, dataset) -- resolve into downloadable chunks
• Chunk -- a single downloadable file
  • fetch(chunk, filepath) -- download to disk
  • filesize(chunk) -- remote file size in bytes (via HEAD request), or nothing
• Project -- spatial/temporal region of interest + output directory

Sources

Source	Description	Auth
NOMADS	NOAA operational weather models (GFS, HRRR, NAM, ...) via GRIB filter	None
CDS	Copernicus Climate Data Store (ERA5, CERRA, satellite, ...)	CDSAPI_KEY env var or ~/.cdsapirc
FIRMS	NASA active fire data (VIIRS, MODIS, LANDSAT)	FIRMS_MAP_KEY env var
ETOPO	NOAA global relief model (bathymetry + topography)	None
SRTM	NASA Shuttle Radar Topography Mission (30m/90m elevation)	EARTHDATA_TOKEN env var
GOES	NOAA GOES-16/17/18 satellite imagery via AWS S3	None
HRRRArchive	HRRR model archive via AWS S3	None
NASAPower	NASA global daily meteorological and solar data	None
USGSWater	USGS streamflow, gage height, and water quality observations	None
NCEI	NOAA historical weather and climate observations	None
OISST	NOAA daily global sea surface temperature (0.25° grid)	None
Landfire	USGS LANDFIRE wildland fire, vegetation, and fuel data via WCS	None
NDBC	NOAA National Data Buoy Center ocean/meteorological observations	None

Project

Project(;
    geometry = EARTH,           # any GeoInterface-compatible geometry or Extent
    datetimes = nothing,        # (start, stop) DateTime tuple
    path = mktempdir(),         # output directory
    datasets = [],              # Vector of Dataset instances
)

• fetch(project) downloads all chunks, skipping files that already exist.
• Data is written to "$(project.path)/data/".