Oregon Vineyard Multispectral & LiDAR — PCC Partnership
Overview
An ongoing multi-sensor aerial survey of an Oregon vineyard conducted in partnership with Portland Community College. The original goal was to produce a high-resolution orthomosaic and Normalized Difference Vegetation Index (NDVI) map of the approximately six-acre site, with LiDAR point cloud data supporting structural and canopy analysis.
The two 3D Gaussian splats (above and below) are reconstructed from imagery collected during the August 2025 campaign. This new method was explored to solve a problem I encountered generating 3D vineyard models with traditional Structure From Motion (SfM) workflows from nadir imagery: the tendency of narrow objects to snap to ground — narrow objects like powerlines in infrastructure GIS work, or in this case narrow Pinot Noir vineyard rows.
Platform & Sensors
All flights were flown nadir in a lawnmower pattern using a DJI Matrice 300 RTK carrying three payloads across the campaign:
- Sentera 6X — six-band multispectral camera capturing Red, Green, Blue, Red Edge, Near-Infrared, and RGB. Primary output for vegetation health analysis.
- Zenmuse L1 — integrated LiDAR and RGB sensor. Produces dense point clouds for canopy height modeling and terrain analysis beneath vine rows.
- DJI P1 — 45MP full-frame photogrammetry camera. High-resolution imagery for orthomosaic production and the Gaussian splat reconstruction.
Methodology
Survey design followed standard agricultural remote sensing protocols: nadir flight lines at consistent altitude with sufficient overlap for photogrammetric processing. Ground control points established for spatial accuracy.
Multispectral data is processed to surface reflectance for various index calculations. LiDAR returns are classified to isolate vine and canopy returns. Photogrammetric outputs feed both the orthomosaic and the 3DGS reconstruction.
Two indices were focused on for Pinot Noir’s specific characteristics: GNDVI (Green Normalized Vegetation Index) for its particular sensitivity to chlorophyll levels and ability to convey overall canopy health, and RECI (Red Edge Chlorophyll Index) to detect signs of chlorophyll level changes and veraison, or ripening, that occur within the leaf structure before presenting in the clusters. These are then cross-referenced to highlight areas of particular concern.
Outputs
- High-resolution orthomosaic, DEM
- GNDVI, RECI map for vine vigor assessment
- LiDAR-derived canopy height / vine structure model and DSM, DEM
- 3D Gaussian splat reconstruction (viewable above)
Project ongoing — additional flight campaigns and seasonal comparisons planned.