Skip to content

Tutorial: NEON -> Landsat-Style Reflectance

Canonical learning path

This older tutorial is retained for stable links. Use Harmonize target sensors for the canonical module vignette.

This tutorial shows how SpectralBridge turns physically corrected NEON ENVI reflectance into Landsat-style sensor products using sensor spectral response functions (SRFs).

The output is a set of ENVI cubes and Parquet tables representing NEON reflectance in target sensor bandspaces such as Landsat TM, ETM+, OLI, and OLI-2.

Prerequisites

Before running this tutorial, complete:

You will need the *_brdfandtopo_corrected_envi.img product. If it already exists and validates, the pipeline will reuse it and continue to sensor resampling.

1. Select a corrected ENVI cube

Your corrected data should live inside a per-flightline folder:

output_neon_to_envi/
└── NEON_D13_NIWO_DP1_L020-1_20230815_directional_reflectance/
    └── NEON_D13_NIWO_DP1_L020-1_20230815_directional_reflectance_brdfandtopo_corrected_envi.img

This corrected ENVI product is the input to sensor bandpass convolution.

2. Run the pipeline

Sensor resampling runs automatically after BRDF + topographic correction. To generate or refresh Landsat-style products, re-run the standard pipeline for the same flight line. Completed earlier stages are skipped.

spectralbridge-pipeline \
  --base-folder "$BASE" \
  --site-code NIWO \
  --year-month 2023-08 \
  --product-code DP1.30006.001 \
  --flight-lines NEON_D13_NIWO_DP1_L020-1_20230815_directional_reflectance \
  --max-workers 2 \
  --engine thread

Outputs include sensor-specific files such as:

*_landsat_tm_envi.img
*_landsat_tm_envi.hdr
*_landsat_tm_envi.parquet
*_landsat_oli_envi.img
*_landsat_oli_envi.hdr
*_landsat_oli_envi.parquet

3. Understanding Landsat bandpass integration

The convolution stage integrates each corrected NEON spectrum against the target sensor SRFs. For Landsat OLI-style products, the output bands include coastal/aerosol, blue, green, red, NIR, SWIR1, and SWIR2 where supported by the sensor definition.

This process produces reflectance values aligned to Landsat definitions, enabling:

  • direct comparison to satellite imagery
  • validation exercises
  • cross-scale modeling
  • harmonized vegetation index calculations

4. Inspecting Landsat-style ENVI outputs

Example in Python:

import rioxarray as rxr

landsat = rxr.open_rasterio("path/to/..._landsat_oli_envi.img")
landsat

Band order, wavelengths, and metadata are stored in the matching ENVI header.

5. Checking QA outputs

The QA JSON and QA PNG summarize:

  • brightness adjustments used in harmonization
  • bandwise statistics after convolution
  • wavelength alignment checks
  • mask and no-data diagnostics

Use these artifacts to verify successful harmonization before downstream analysis.

6. Using the Parquet tables

Sensor Parquet files use one row per pixel-band observation. A quick DuckDB summary looks like:

import duckdb

duckdb.query("""
    SELECT wavelength_nm, AVG(reflectance) AS mean_reflectance
    FROM '..._landsat_oli_envi.parquet'
    GROUP BY wavelength_nm
    ORDER BY wavelength_nm
""").df()

For analysis across all products, start with the merged table: *_merged_pixel_extraction.parquet.

Next steps