Small area estimation of tree diameter distributions under a multivariate compositional Fay–Herriot model
Forestry: An International Journal of Forest Research · 26.06.2026 · 99 (4)
DOI: 10.1093/forestry/cpag051 ↗Kurz gefasst
Fachartikel in der Zeitschrift Forestry: An International Journal of Forest Research, online erschienen am 29.07.2026. Verfasst von Aristeidis Georgakis und 3 weiteren Autor:innen. Die Arbeit nennt 62 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Accurate estimation of tree diameter distributions is essential for sustainable forest management but remains challenging when most forest management units (FMUs) contain only a few sample plots. Standard parametric models (e.g. Weibull) require large within-unit samples or many tens to hundreds of trees per stand to achieve reliable fits and therefore perform poorly in data-sparse inventories. We extend the area-level multivariate Fay–Herriot model to a trivariate compositional mixed model (TriCompFH) for small-area estimation of tree diameter-class distributions. The new methodology (i) applies an additive log-ratio transformation to satisfy the unit-sum constraint of compositional data, (ii) jointly models three diameter at breast height (DBH)–class proportions while deriving the fourth as the complement, and (iii) integrates auxiliary canopy-height metrics from WorldView-2 imagery with historical census data. Random effects are estimated via residual maximum likelihood, and uncertainty is quantified using a parametric bootstrap. TriCompFH was applied to clustered FMUs in a heterogeneous uneven-aged fir forest dominated by Abies borisii-regis Mattf. Model diagnostics confirmed normality and homoscedasticity, and comparisons with design-based estimators showed close agreement on average. TriCompFH reduced relative uncertainty by 68%–77% across the three modelled DBH classes, with per-small-area coefficients of variation <5% in the majority of the small areas—a substantial improvement over direct estimators. Beyond class proportions, TriCompFH provides stem densities and stand indices (stand density index, relative density index, q-ratios, recruitment efficiency index), reproducing the inverse J-shaped structure typical of uneven-aged forests while highlighting recruitment bottlenecks and over-stocked domains. TriCompFH delivers reliable DBH-class estimates for data-sparse inventories, enabling informed, density-based management decisions.
Zitieren
APA
Georgakis, A., Papageorgiou, V. E., Diamantopoulou, M. J., & Stamatellos, G. (2026). Small area estimation of tree diameter distributions under a multivariate compositional Fay–Herriot model. Forestry: An International Journal of Forest Research, 99(4). https://doi.org/10.1093/forestry/cpag051
Vancouver
Georgakis A, Papageorgiou VE, Diamantopoulou MJ, Stamatellos G. Small area estimation of tree diameter distributions under a multivariate compositional Fay–Herriot model. Forestry: An International Journal of Forest Research. 2026;99(4). doi:10.1093/forestry/cpag051
BibTeX
@article{Georgakis2026,
author = {Georgakis, Aristeidis and Papageorgiou, Vasileios E and Diamantopoulou, Maria J and Stamatellos, Georgios},
title = {Small area estimation of tree diameter distributions under a multivariate compositional Fay–Herriot model},
journal = {Forestry: An International Journal of Forest Research},
year = {2026},
volume = {99},
number = {4},
doi = {10.1093/forestry/cpag051}
}
EUVIMED