sen2nbar¶

Nadir BRDF Adjusted Reflectance (NBAR) for Sentinel-2 in Python

Overview¶

First, a super small glossary:

BRDF: Bidirectional Reflectance Distribution Function.
DN: Digital Number.
NBAR: Nadir BRDF Adjusted Reflectance.
SR: Surface Reflectance.
STAC: SpatioTemporal Assets Catalogs.

Second, the amazing bibliography by David P. Roy et al., used to create this package:

Multi-temporal MODIS-Landsat data fusion for relative radiometric normalization, gap filling, and prediction of Landsat data.
A general method to normalize Landsat reflectance data to nadir BRDF adjusted reflectance.
Examination of Sentinel-2A multi-spectral instrument (MSI) reflectance anisotropy and the suitability of a general method to normalize MSI reflectance to nadir BRDF adjusted reflectance.
Adjustment of sentinel-2 multi-spectral instrument (MSI) red-edge band reflectance to nadir BRDF adjusted reflectance (NBAR) and quantification of red-edge band BRDF effects.

Third, the super useful bibliography by Lucht et al.,:

An algorithm for the retrieval of albedo from space using semiempirical BRDF models.

Given this, and in a few words, sen2nbar converts the Sentinel-2 SR (i.e., L2A) to Sentinel-2 NBAR via the _c_-factor method.

SAFE¶

You can use sen2nbar to convert complete images via SAFE:

from sen2nbar.nbar import nbar_SAFE

# Converted images are saved inside the SAFE path
nbar_SAFE("S2A_MSIL2A_20230223T075931_N0509_R035_T35HLC_20230223T120656.SAFE")

Note

Note that sen2nbar automatically shifts the DN of images with a processing baseline >= 04.00. This includes data cubes obtained via stackstac or cubo.

stackstac¶

Or, if you are using STAC and retrieving images via stackstac:

import pystac_client
import stackstac
import planetary_computer as pc
from sen2nbar.nbar import nbar_stackstac

# Important infor for later
endpoint = "https://planetarycomputer.microsoft.com/api/stac/v1"
collection = "sentinel-2-l2a"
bounds = (-148.565368, 60.800723, -147.443389, 61.183638)

# Open the STAC
catalog = pystac_client.Client.open(endpoint, modifier=pc.sign_inplace)

# Define your area
area_of_interest = {
   "type": "Polygon",
   "coordinates": [
      [
            [bounds[0], bounds[1]],
            [bounds[2], bounds[1]],
            [bounds[2], bounds[3]],
            [bounds[0], bounds[3]],
            [bounds[0], bounds[1]],
      ]
   ],
}

# Search the items
items = catalog.search(
   collections=[collection],
   intersects=area_of_interest,
   datetime="2019-06-01/2019-08-01",
   query={"eo:cloud_cover": {"lt": 10}},
).get_all_items()

# Retrieve all items as a xr.DataArray
stack = stackstac.stack(
   items,
   assets=["B05","B06","B07"], # Red Edge here, but you can use more!
   bounds_latlon=bounds,
   resolution=20
)

# Convert it to NBAR!
da = nbar_stackstac(
   stack,
   stac=endpoint,
   collection=collection
)

Warning

These examples are done using Planetary Computer. If you are using data cubes retrieved via STAC (e.g., by using stackstac or cubo), we recommend you to use this provider. The provider Element84 is not supported at the moment.

cubo¶

And going deeper, if you are using cubo:

import cubo
import xarray as xr
from sen2nbar.nbar import nbar_cubo

# Get your cube
da = cubo.create(
   lat=47.84815,
   lon=13.37949,
   collection="sentinel-2-l2a",
   bands=["B02","B03","B04"], # RGB here, but you can add more bands!
   start_date="2020-01-01",
   end_date="2021-01-01",
   edge_size=64,
   resolution=10,
   query={"eo:cloud_cover": {"lt": 3}}
)

# Convert it to NBAR (This a xr.DataArray)
da = nbar_cubo(da)

Bands¶

sen2nbar converts the following bands (if available in the input data):

RGB Bands: 02, 03, 04.
Red Edge Bands: 05, 06, 07.
Broad NIR Band: 08.
SWIR Bands: 11, 12.

Installation¶

Install the latest version from PyPI:

pip install sen2nbar

Upgrade sen2nbar by running:

pip install -U sen2nbar

Install the latest version from conda-forge:

conda install -c conda-forge sen2nbar

Install the latest dev version from GitHub by running:

pip install git+https://github.com/davemlz/sen2nbar

License¶

The project is licensed under the MIT license.