Products

From raw counts to indices

Every level is available, because someone always needs the one below the one you thought was enough.

Processing levels

Level	Description	Format	Typical size
L1C	Top-of-atmosphere reflectance, orthorectified	COG, 12-bit	780 MB
L2A	Surface reflectance with cloud and shadow masks	COG, 12-bit	812 MB
L2A-mosaic	Cloud-free monthly composite per tile	COG, 12-bit	640 MB
Indices	NDVI, NDWI, NBR, EVI2 and 14 more	COG, float32	68 MB each
SAR GRD	Terrain-corrected backscatter, γ⁰	COG, float32	1.6 GB
SAR coherence	Interferometric coherence, 6/12-day pairs	COG, float32	2.1 GB

Derived on request

Time series extraction

Point or polygon, any index, any date range, returned as Parquet or CSV without you touching a raster.

Custom composites

Median, geometric median or best-pixel composites over any window, produced once and cached.

Change detection

Pairwise or trend-based change layers with the algorithm and thresholds recorded in the item metadata.

Reading a window without downloading a scene

window.py

import rasterio
from rasterio.windows import from_bounds

url = (&#x27;https://data.mirnl.13381338.xyz/s2/32/U/NB/'
       &#x27;2026/08/02/L2A/B08.tif')

with rasterio.open(url) as src:
    win = from_bounds(410000, 5650000, 412000, 5652000,
                      src.transform)
    nir = src.read(1, window=win)   # 4 MB moved, not 812

print(nir.shape, nir.mean())