Skip to content

Why cross-sensor calibration?

Understanding vegetation, disturbance, and ecosystem structure often requires integrating information from multiple sensors operating at different spatial, temporal, and spectral scales. These include:

  • NEON airborne imaging spectroscopy
  • drone multispectral systems (e.g., MicaSense)
  • moderate-resolution satellites like Landsat and Sentinel

Each sensor “sees” the landscape differently, and these differences can obscure the ecological signals we care about unless we account for them.

The problem: apples and oranges reflectance

Sensors differ in:

  • spectral response (band centers, widths, shapes)
  • illumination geometry (solar zenith, azimuth, atmospheric path)
  • viewing geometry (sensor zenith, azimuth)
  • radiometric scaling and masking conventions
  • ground sampling distance and spatial aggregation

Even when they image the same location on the same day, their raw reflectance values are not directly comparable.

This creates challenges when trying to:

  • validate satellite products using NEON
  • relate drone measurements to NEON or Landsat
  • build cross-scale ecological models
  • interpret changes in reflectance through time or across terrain

Correcting vs. harmonizing

cross-sensor-cal performs two distinct operations:

1. Physical corrections

These aim to reduce variation caused by illumination and terrain:

  • topographic correction (slope and aspect effects)
  • BRDF correction (view/sun geometry effects)

The result is a reflectance product that is more comparable across acquisition conditions.

2. Sensor harmonization

This converts corrected hyperspectral data into another sensor’s bandspace by integrating spectra against published spectral response functions (e.g., Landsat OLI, MicaSense RedEdge).

Optional brightness adjustments are documented in the QA outputs.

Why NEON as the foundation?

NEON AOP data provide:

  • high spectral resolution
  • per-pixel geometry information
  • consistent radiometric processing
  • spatial coverage aligned with ecological research sites

These properties make NEON a powerful intermediary between plot-scale measurements and satellite observations.

cross-sensor-cal implements a reproducible stepwise process to:

  1. extract NEON reflectance into ENVI
  2. correct it physically
  3. harmonize it to satellite/drone sensors
  4. output analysis-ready tables and QA documentation

What still requires care

Even after calibration and harmonization:

  • residual BRDF effects can remain
  • atmospheric differences between sensors matter
  • snow, smoke, water, and shadows require attention
  • scale mismatch affects interpretation
  • masks and quality flags differ across platforms

The pipeline aims to make all assumptions explicit—every major processing step writes a JSON sidecar describing inputs, parameters, and results.

Where to go next