EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

A Bayesian framework for mapping subtropical montane cloud forest litterfall and its uncertainty from airborne LiDAR canopy structure

Hsin-He Chang, Chih-Hsin Chung, Chi-Kuei Wang, Cho-Ying Huang

Science of Remote Sensing · 2026-12 · 14 · 100545

DOI: 10.1016/j.srs.2026.100545 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science of Remote Sensing, 2026-12. Verfasst von Hsin-He Chang und 3 weiteren Autor:innen. Die Arbeit nennt 83 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: National Science and Technology Council.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Litterfall is a major pathway of forest carbon cycling, but field monitoring is labor-intensive for regional application. We developed a framework to map annual litterfall flux and its pixel-level uncertainty in subtropical montane cloud forests of northeastern Taiwan. The framework combined 2021–2023 annual litterfall flux and mean monthly leaf area index (LAI) from 16 plots (n = 48, 16 plots × 3 years) with 1-m airborne LiDAR canopy height models (CHMs) acquired in 2019, 2021, and 2024. Annual CHMs were projected across the acquisition gaps using an empirical canopy height growth model, enabling regional LAI upscaling (R 2 = 0.77 for both cross- and external validation). A hierarchical Bayesian model (Gamma likelihood, log link) related litterfall to LAI-derived leaf mass through a leaf-mass coefficient, summarized as an apparent litterfall-to-leaf-mass ratio (1.33 yr -1 ), and propagated uncertainty to every map pixel. Litterfall was mapped as the posterior predictive median (LF P50 ) with pixel-wise 95% credible intervals; regional medians were 4.59, 3.35, and 3.71 Mg ha -1 yr -1 in 2021–2023. Spatial heterogeneity was associated with canopy structure and, in a post hoc analysis, terrain aspect. In an exploratory analysis, replacing the year-level random effect of the leaf-mass coefficient with an annual radiation anomaly from global reanalysis data left the predictions nearly unchanged. The three-year record precludes causal inference, but the hierarchical structure allows this hypothesis to be re-evaluated as data accumulate. This study demonstrates how multi-temporal airborne LiDAR and Bayesian modeling can link plot-scale measurements to regional carbon dynamics in topographically complex forests.

Förderung

  • National Science and Technology Council

Zitieren

APA

Chang, H., Chung, C., Wang, C., & Huang, C. (2026). A Bayesian framework for mapping subtropical montane cloud forest litterfall and its uncertainty from airborne LiDAR canopy structure. Science of Remote Sensing, 14, 100545. https://doi.org/10.1016/j.srs.2026.100545

Vancouver

Chang H, Chung C, Wang C, Huang C. A Bayesian framework for mapping subtropical montane cloud forest litterfall and its uncertainty from airborne LiDAR canopy structure. Science of Remote Sensing. 2026;14:100545. doi:10.1016/j.srs.2026.100545

BibTeX
@article{Chang2026,
  author = {Chang, Hsin-He and Chung, Chih-Hsin and Wang, Chi-Kuei and Huang, Cho-Ying},
  title = {A Bayesian framework for mapping subtropical montane cloud forest litterfall and its uncertainty from airborne LiDAR canopy structure},
  journal = {Science of Remote Sensing},
  year = {2026},
  volume = {14},
  pages = {100545},
  doi = {10.1016/j.srs.2026.100545}
}
Kostenlos starten →