Published September 2, 2026 | Version v1

Data from: Using a planted tree biodiversity experiment to evaluate imaging spectroscopy for species classification

  • 1. University of Zurich
  • 2. University of Freiburg
  • 3. Forschungszentrum Jülich

Description

This dataset contains airborne imaging spectroscopy measurements of tree canopies collected at two sites of the BIOTREE tree biodiversity experiment in Germany (Kaltenborn and Bechstedt). The experiment comprises monospecific tree patches planted under controlled diversity levels on relatively homogeneous plots, reducing confounding effects of habitat preferences, tree age, crown overlap, and environmental heterogeneity. Reflectance data were collected during peak growing season at 1 m spatial resolution across 589 wavelengths between 400 and 2400 nm. Analyses primarily used the three brightest pixels per patch to focus on sunlit canopy elements, with additional analyses using the ten central pixels to include shaded and illuminated pixels. Results were generally similar between approaches. Reflectance spectra differed significantly among taxonomic groups and species. Classification accuracy varied with species number, reaching 77–83 % for four species at Kaltenborn and 31–49 % for 16 species at Bechstedt; LDA generally outperformed SVM. The dataset therefore provides controlled hyperspectral measurements for investigating spectral variation among tree species and diversity levels.

Notes

Funding provided by: Nomis Foundation
ROR ID: https://ror.org/01zkd3118
Award Number:

Funding provided by: Swiss National Science Foundation
ROR ID: https://ror.org/00yjd3n13
Award Number:

Methods

Data were collected on 17 June 2021 at the BIOTREE biodiversity experiment sites in Kaltenborn and Bechstedt, Germany, using the DUAL module of the airborne HyPlant imaging spectrometer. Four flight lines at Kaltenborn and two at Bechstedt were acquired at approximately 680 m above ground, yielding a spatial resolution of about 1 m. Radiance data were radiometrically and geometrically corrected and converted to top-of-canopy reflectance using ATCOR-4. The resulting data comprise 589 spectral bands covering 400–2500 nm.

Reflectance pixels were assigned to monospecific tree patches using the experimental planting maps. To reduce edge effects and shading, analyses primarily used the three brightest of the ten most central pixels within each patch; mean reflectance was calculated for each patch. The dataset includes measurements from the two experimental sites, representing different tree species and experimentally manipulated species or functional diversity levels.

Files

Files (143.5 kB)

Name Size
md5:bbfdab14c2a9546890440ee3eaf90428
122.2 kB Download
md5:9766d831c42fad6b8b0b2f4723a548cd
21.3 kB Download

Additional details

Related works

Is cited by
10.64898/2026.03.20.713086 (DOI)
Is source of
10.5061/dryad.2bvq83c60 (DOI)