gci

gci(nir, green)[source]

Green Chlorophyll Index (GCI)

GCI = (NIR / Green) - 1

Parameters:

  • nir (numpy.ndarray) – Near-infrared band (any numeric dtype; auto-coerced to float64).

  • green (numpy.ndarray) – Green band (any numeric dtype).

Returns:

GCI values (unbounded; typical vegetation > 0).

Return type:

numpy.ndarray

Notes

Division-by-near-zero guarded; returns 0 where Green is ~0.

Overview

The Green Chlorophyll Index (GCI) is a simple reflectance‑based proxy for chlorophyll content and plant vigor. It uses the ratio between near‑infrared (NIR) and green bands:

Formula

\[\mathrm{GCI} = \frac{\mathrm{NIR}}{\mathrm{Green}} - 1\]

Where: - NIR = near‑infrared reflectance (typically 0–1 reflectance units) - Green = green band reflectance

A near‑zero green reflectance is guarded internally; the implementation returns 0.0 when the denominator is below an epsilon threshold to avoid numerical instability.

Typical Interpretation

  • GCI > 0 generally indicates presence of vegetation.

  • Larger positive values can correlate with higher chlorophyll concentration (exact biophysical mapping requires calibration against field measurements or radiative transfer modeling).

Usage Example

import numpy as np
from eo_processor import gci

nir   = np.array([0.80, 0.90, 0.70], dtype=np.float64)
green = np.array([0.40, 0.45, 0.35], dtype=np.float64)

out = gci(nir, green)
print(out)          # (nir/green) - 1 element-wise

Dimensions & Dtypes

  • Supports 1D and 2D arrays (the function internally dispatches to matching loops; higher ranks may be supported indirectly via shared normalized difference logic but are not part of the documented contract here).

  • Inputs may be any numeric dtype; values are coerced to float64 for stable arithmetic.

  • Output array is float64.

Edge Cases

  • Green ≈ 0: Result forced to 0.0 (instead of exploding toward ±∞).

  • Shape mismatch between nir and green raises a ValueError.

  • Non 1D/2D input raises a TypeError.

Performance Notes

Loop executed in Rust (PyO3) with: - Early shape validation (clear error rather than silent broadcast) - Single fused pass (no temporaries) - Float64 coercion once per input

See Also

References

  • Concept popularized in remote sensing literature as a simple reflectance ratio derivative. For rigorous chlorophyll retrieval consider PROSPECT + SAIL model inversions or empirical regression against leaf samples.

End of GCI documentation.