Skip to content

Application — vegetation greenness

Goal: Run one application that uses every workshop tool: STAC, COG, xarray, rioxarray, and Dask.

Question: How does Sentinel-2 NDVI change from January to June 2023 over a small AOI in northern Karnataka, and is greenness different on higher ground?

Pipeline

STAC search (S2 + DEM)  →  sign URLs
       →  open_rasterio(url, chunks)  →  clip_box
       →  reproject_match (DEM onto S2)
       →  concat time  →  NDVI
       →  persist / mean / mask  →  plot + GeoTIFF / NetCDF

Clip first. Data stays in the cloud until a plot or write.

What each lesson does here

Application step Tool Lesson
Search S2 + DEM by bbox, time, cloud cover STAC + planetary_computer.sign 1
Open a COG; read CRS / shape only COG + rioxarray 1
Clip the window; warp DEM onto S2 .rio.clip_box, .rio.reproject_match 3
Stack dates; NDVI; .sel / .isel xarray Dataset 2
chunks=, persist, spatial mean, elevation mask Dask 4
RGB + NDVI maps, time series, write matplotlib, .rio.to_raster 1–4

Search (STAC)

CLIP = (76.40, 16.75, 76.55, 16.90)

catalog = pystac_client.Client.open(
    "https://planetarycomputer.microsoft.com/api/stac/v1"
)
search = catalog.search(
    collections=["sentinel-2-l2a"],
    bbox=list(CLIP),
    datetime="2023-01-01/2023-06-30",
    max_items=5,
    query={"eo:cloud_cover": {"lt": 15}},
)
items = [planetary_computer.sign(it) for it in search.items()]

A Copernicus DEM tile for the same box is signed the same way (cop-dem-glo-30).

Cube + NDVI (rioxarray, xarray, Dask)

Open B04 and B08 with chunks={"x": 1024, "y": 1024}, clip_box to CLIP, concat along time, then:

cube["ndvi"] = (cube["B08"] - cube["B04"]) / (cube["B08"] + cube["B04"])
ts = cube["ndvi"].mean(dim=("y", "x")).compute()
ndvi_high = cube["ndvi"].where(cube["dem"] > 500).mean(dim=("y", "x")).compute()

The DEM is aligned with reproject_match onto the first Sentinel-2 grid (Lesson 3).

Outputs

  • RGB preview from the visual asset
  • NDVI map for the first date
  • Time series: AOI mean vs DEM > 500 m
  • output/ndvi_YYYY-MM-DD.tif, dem_on_s2.tif, karnataka_ndvi_dem.nc

Download the notebook for this lesson: notebooks/05-end-to-end.ipynb.