nbr¶

nbr(nir, swir2)[source]¶

Compute Normalized Burn Ratio (NBR).

NBR = (NIR - SWIR2) / (NIR + SWIR2)

Parameters:

nir (numpy.ndarray) – Near-infrared band.
swir2 (numpy.ndarray) – Short-wave infrared (SWIR2) band.

Returns:

NBR values with same shape as inputs.

Return type:

numpy.ndarray

Overview¶

nbr computes the Normalized Burn Ratio (NBR):

\[\mathrm{NBR} = \frac{NIR - SWIR2}{NIR + SWIR2}\]

This index is widely used for burn severity assessment. Comparing pre‑ and post‑event NBR via delta_nbr highlights fire impacts (larger positive Δ often indicates higher severity).

Parameters¶

nirnumpy.ndarray: Near‑infrared reflectance band (any numeric dtype; internally coerced to float64).
swir2numpy.ndarray: Short‑wave infrared 2 reflectance band (same shape & dtype category as nir).

Returns¶

numpy.ndarray: NBR values with the same shape as inputs (float64). Values typically range from -1 to 1.

Numeric Stability¶

For pixels where nir + swir2 is ~0, an epsilon guard returns 0.0 to avoid division instability. This prevents spurious large magnitude values when both bands are near zero.

Shape Support¶

1D: (n_pixels,)
2D: (rows, cols)
3D / 4D: Supported internally via the generic normalized difference primitive if higher‑rank arrays are passed (e.g., (time, y, x) or (time, band, y, x)). Public documentation emphasizes 1D/2D usage; ensure matching shapes between nir and swir2 for all dimensions.

Typical Interpretation (Approximate)¶

High positive values: Recently burned regions often show reduced NIR and increased SWIR reflectance.
Moderate: Mixed or partially recovered vegetation.
Low / negative: Water, snow, or non‑vegetated surfaces.

Example¶

import numpy as np
from eo_processor import nbr

nir   = np.array([0.6, 0.5, 0.4], dtype=np.float32)
swir2 = np.array([0.3, 0.2, 0.1], dtype=np.float32)
out = nbr(nir, swir2)
print(out)

End of NBR reference.