Skip to content

Module vignette 6: process drone imagery

Notebook: View the drone notebook in the repository. GitHub displays the cells; clone or download the file to run them.

Use this module for local drone HDF5 inputs. It is separate from the NEON download workflow and preserves provenance from the original drone filenames.

Where Landsat enters the scientific bridge

The project-level relationship is Drone → NEON → Landsat. This module prepares and validates the drone side; it does not directly convolve drone data into Landsat bands. Corrected, Landsat-harmonized NEON products provide the explicit airborne translation reference for downstream comparison.

Prepare the inputs

Place valid drone HDF5 files under one input directory. Discovery is recursive, so campaign or flight subdirectories are allowed.

drone_inputs/
├── flight_a/
│   └── flight_a.h5
└── flight_b/
    └── flight_b.h5

Reflectance and ancillary arrays in each HDF5 must already share the expected spatial orientation and footprint. Corrections require the relevant terrain, view, and solar geometry.

Run it

from spectralbridge import run_drone_pipeline

results = run_drone_pipeline(
    input_h5_dir="drone_inputs",
    output_dir="drone_outputs",
    polygon_path=None,
    apply_topo=True,
    apply_brdf=True,
    require_solar_geometry=True,
)

print(results["processed"])
print(results["failed"])
print(results["qa_summary"])

Set polygon_path to a supported vector file only when polygon extraction is part of the same run.

Confirm the result

Each discovered flight gets a separate output directory with drone-native stems. Depending on available inputs and requested modules, expect corrected ENVI, Parquet, QA PNG, and QA JSON outputs plus a run-level QA summary.

Read each QA JSON before assuming a requested correction was applied. It records whether correction ran, was skipped because geometry was unavailable, or fell back after a quality check.

Continue