EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ForestForTrees : An R package to infer forest structure from remote sensing data

Adam J. Eichenwald, John M. Grady, Alyson East, Noah Charney, Sydne Record

Methods in Ecology and Evolution · 2026-07 · 17 (7) · 2052-2060

DOI: 10.1111/2041-210x.70309 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Methods in Ecology and Evolution, online erschienen am 21.05.2026. Verfasst von Adam J. Eichenwald und 4 weiteren Autor:innen. Die Arbeit nennt 20 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: U.S. Department of Agriculture, National Institute of Food and Agriculture, National Aeronautics and Space Administration.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Estimating forest size‐abundance distributions is essential for understanding demographic processes and structural dynamics. Remote sensing provides broad coverage but often misses smaller trees hidden beneath the canopy, complicating inference of the full size distribution. Although Bayesian scaling frameworks exist to correct for this occlusion, their application has previously required advanced proficiency in probabilistic programming and manual, site‐specific tuning, preventing broad adoption for large‐scale monitoring or management. We present ForestForTrees , an R package that operationalizes this framework into a generalized, accessible and fully reproducible pipeline. ForestForTrees makes three distinct contributions beyond previously published methods. First, it automates Markov Chain Monte Carlo sampling, uncertainty propagation and observational window detection into a sequential pipeline compatible with any upstream segmentation method or sensor type, requiring only a vector of estimated tree diameters and a site‐level leaf area index value as inputs. Second, it introduces ‘map_grid_estimates()’, a new function that joins plot‐level estimates of the Pareto shape parameter ( α ) and total tree abundance ( N tot ) to spatial polygon layers, enabling landscape‐scale raster mapping of forest structural parameters without additional scripting. Third, it includes a novel cross‐biome computational benchmark analysis across 20 National Ecological Observatory Network (NEON) sites spanning subtropical to alpine forest types, demonstrating robust parameter recovery and reporting runtime performance to help users assess the practicality of large‐scale deployment. We demonstrate the package through two complementary empirical case studies. The benchmarking analysis across 20 NEON sites confirms that the package recovers both α ( R 2 = 0.54) and total tree abundance ( R 2 = 0.61) with no systematic bias and without site‐specific reconfiguration. A step‐by‐step workflow demonstration using Harvard Forest data illustrates the full preprocessing pipeline and the gridded mapping functionality, with a fully reproducible vignette provided for each case study. ForestForTrees enables transparent, sensor‐agnostic inference of complete forest size‐abundance distributions from remotely sensed canopy data, supporting applications from ecological research and habitat assessment to forest management and global carbon monitoring.

Förderung

  • U.S. Department of Agriculture
  • National Institute of Food and Agriculture
  • National Aeronautics and Space Administration

Zitieren

APA

Eichenwald, A. J., Grady, J. M., East, A., Charney, N., & Record, S. (2026). ForestForTrees : An R package to infer forest structure from remote sensing data. Methods in Ecology and Evolution, 17(7), 2052-2060. https://doi.org/10.1111/2041-210x.70309

Vancouver

Eichenwald AJ, Grady JM, East A, Charney N, Record S. ForestForTrees : An R package to infer forest structure from remote sensing data. Methods in Ecology and Evolution. 2026;17(7):2052-2060. doi:10.1111/2041-210x.70309

BibTeX
@article{Eichenwald2026,
  author = {Eichenwald, Adam J. and Grady, John M. and East, Alyson and Charney, Noah and Record, Sydne},
  title = {ForestForTrees : An R package to infer forest structure from remote sensing data},
  journal = {Methods in Ecology and Evolution},
  year = {2026},
  volume = {17},
  number = {7},
  pages = {2052--2060},
  doi = {10.1111/2041-210x.70309}
}
Kostenlos starten →