Skip to main content

NASA-NEX GDDP-CMIP6 Downloader

NASA-NEX GDDP-CMIP6 Downloader​

NasaNexDownloader fetches NASA-NEX GDDP-CMIP6 downscaled climate-projection granules for any bounding-box AOI via the Microsoft Planetary Computer STAC API and stages the NetCDF-4 granules into a Unity Catalog Volume.

NASA-NEX GDDP-CMIP6 provides daily downscaled climate variables (temperature, precipitation, humidity, radiation, …) on a regular 0.25° global lat/lon grid, one file per climate model, scenario, and year. Because these are true georeferenced grids — not swaths — they pair with the NetCDF reader's raster mode, and can be read by both the lightweight netcdf_gbx and the heavyweight netcdf_gdal readers.

Prerequisites
  • GeoBrix installed (wheel includes databricks.labs.gbx.sample)
  • Unity Catalog Volume already exists at /Volumes/{catalog}/{schema}/{volume}/...
  • pystac-client and planetary-computer packages installed: %pip install pystac-client planetary-computer
  • No authentication required — NASA-NEX GDDP-CMIP6 is public on Planetary Computer

How It Works​

NasaNexDownloader follows the same discover → download → read pattern as the other Planetary Computer downloaders, staging regular-grid NetCDF granules:

  1. discover(bbox, temporal=None, variables=("tas",)) — driver-side STAC search against the nasa-nex-gddp-cmip6 collection. Returns one row per matching granule: item_id, asset_name, item_bbox, href. In this collection the asset name is the climate-variable id (tas, pr, tasmax, hurs, …), so variables selects which variables to stage. A granule spans one model/scenario/year; narrow the set with a temporal STAC-style "start/end" string.

  2. download(bbox, out_dir, temporal=None, variables=("tas",), ...) — re-runs the STAC search, then fans out the granule fetch as parallel Spark tasks via StacClient.download(), saving each file as {item_id}_{asset_name}.nc (one file per climate variable per item). Regular-grid granules are fetched whole and existence/size-validated; the AOI clip happens on read. Returns a metadata DataFrame: item_id, asset_name, out_file_path, out_file_sz, is_out_file_valid, last_update.

  3. read(out_dir) — loads the staged .nc granules through the netcdf_gbx reader in raster mode: one (source, tile) row per grid variable, with the source path preserved so multiple variables stay distinguishable.

Example​

from databricks.labs.gbx.sample import NasaNexDownloader, download_nasanex_aoi

# US Southwest AOI, a short window, the near-surface air-temperature variable.
bbox = [-104.5, 31.0, -101.5, 33.0]
out_dir = "/Volumes/main/geobrix_samples/nasanex"

# One-shot convenience wrapper.
manifest = download_nasanex_aoi(spark, bbox, out_dir,
                                temporal="2014-01-01/2014-01-02",
                                variables=("tas",))
manifest.select("item_id", "asset_name", "out_file_sz", "is_out_file_valid").show()

# Or drive the downloader directly, then read the staged grids as raster tiles.
dl = NasaNexDownloader()
dl.download(bbox, out_dir, temporal="2014-01-01/2014-01-02", variables=("tas", "pr"), spark=spark)
grids = dl.read(out_dir)   # (source, tile) rows, raster mode

Each granule is a full-year daily global grid (hundreds of MB); bound the corpus with variables and a short temporal window when staging for interactive work.