Your first dataset

Start with a provider surface-reflectance product and a small spatial subset. This avoids downloading atmospheric models or running a retrieval just to learn the data interface.

1. Install and import

Follow installation to install this checkout with Python >=3.11. If you need an input scene first, use the search and download workflow.

import hyperproc as hp
print(hp.__version__)
print(hp.summary())

2. Open and inspect

ds = hp.open("/path/to/provider_reflectance_product")
hp.describe(ds)
print(ds.attrs)
print(ds.wavelength.values)

hp.open() selects a reader using the filename, or explicit sensor= and level= where supported. The returned cube generally has (y, x, wavelength) dimensions. Opening is lazy for the cube, but reading metadata, geometry checks, and describe() can still access data.

3. Work on a window

small = ds.isel(y=slice(0, 200), x=slice(0, 200))
q = hp.quality_flags(small)
usable = hp.quality_apply(small, q)
ndvi = hp.spectral_index(usable, "NDVI")

Choose a window inside valid observations, not automatically the upper-left corner of a rotated or padded scene. Missing provider flags do not prove clear-sky conditions.

4. Export a mapped product

mapped = usable if usable.attrs.get("crs") else hp.georeference(usable)
hp.to_geotiff(mapped, "products/first_scene.tif")
hp.bands_to_csv(mapped, "products/first_scene_bands.csv")

Export writes files and triggers computation. For a swath without usable latitude/longitude, do not invent a CRS; obtain the required geolocation first.