Tutorial: MicaSense Local-H5 Workflow

This tutorial shows the supported path for local drone or MicaSense-style HDF5 exports. The workflow preserves drone-native provenance, discovers HDF5 files recursively, writes ENVI products, extracts polygon Parquet tables when a polygon layer is provided, and renders QA artifacts.

There is no dedicated spectralbridge-micasense-to-landsat CLI today. Use the Python API shown here.

Inputs

Prepare a folder containing local HDF5 exports:

drone_h5_exports/
  SPR1-06-28-23 ExportPackage/
    *.h5
  SPR2-06-28-23 ExportPackage/
    *.h5

If you want polygon-level tables, also prepare a polygon layer supported by GeoPandas, such as GeoPackage or GeoJSON.

HDF5 contract

SpectralBridge treats the local HDF5 export as the authoritative drone input. Reflectance and ancillary rasters are expected to already share the same spatial orientation and (lines, columns) footprint. SpectralBridge validates that contract during loading and correction, but it does not add TIFF conversion logic or attempt to repair malformed upstream TIFF-to-HDF5 exports.

Run the drone pipeline

from spectralbridge import run_drone_pipeline

results = run_drone_pipeline(
    input_h5_dir="drone_h5_exports",
    polygon_path="Datasets/niwot_aop_polygons_2023_12_8_23_analysis_ready_half_diam.gpkg",
    output_dir="drone_outputs",
    apply_topo=True,
    apply_brdf=True,
    use_ndvi_brdf_bins=False,
    apply_brightness_adjustment=False,
)

Set polygon_path=None when you only need ENVI and QA products.

Outputs

Each discovered flight gets its own output folder:

drone_outputs/
  <flight_stem>/
    <flight_stem>__working.h5
    <flight_stem>__envi.img
    <flight_stem>__envi.hdr
    <flight_stem>__corrected.img
    <flight_stem>__corrected.hdr
    <flight_stem>__polygon_index.parquet
    <flight_stem>__polygons.parquet
    <flight_stem>__qa.png
    <flight_stem>__qa.json
  drone_merged.parquet
  drone_qa_summary.json

Parquet remains the authoritative tabular output. CSV sidecars, when present in drone workflows, are convenience copies for external tools.

Inspect the run

from pathlib import Path

output_dir = Path("drone_outputs")
print(results["qa_summary_path"])
print(results["merged"])

for qa_png in output_dir.glob("*/*__qa.png"):
    print(qa_png)

The QA JSON records whether requested topographic or BRDF corrections were applied, skipped because ancillary data were unavailable, or reverted because the corrected product failed quality checks.

Relation to Landsat harmonization

The local drone path does not invent NEON-style filenames and does not run the NEON download workflow. Landsat-style products for NEON flightlines are still generated by the standard sensor harmonization stage and use names such as:

*_landsat_tm_envi.img
*_landsat_oli_envi.img
*_micasense_envi.img
*_micasense_to_match_oli_oli2_envi.img

Those NEON outputs are documented in Outputs & file structure.