EUVIMEDEUVIMED entdecken
Crossref

Artificial Neural Networks for early prediction of eucalyptus stem diameter across climatic gradientes

William de Medeiros Silva, Márcio José de Araujo, Otavio Camargo Campoe, Rinaldo Cesar de Paula

Scientia Forestalis · 03.06.2026 · 54 · 1-9

DOI: 10.18671/scifor.v54.08 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Scientia Forestalis, online erschienen am 03.06.2026. Verfasst von William de Medeiros Silva und 3 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Developing an artificial neural network (ANN) model for predicting eucalyptus growth where no plantations have been previously established is extremely important. Thus, this work aims to model the growth of a eucalypt clonal stand using ANN and to evaluate the model’s ability to be applied to other sites with different genetic materials and edaphoclimatic conditions. The used data files were obtained from continuous forest inventories of 18 commercial Eucalyptus clones from 17 sites in different states of Brazil, that are part of the TECHS project (Tolerance of Clonal Eucalyptus to Water, Thermal and Biotic Stresses). The multilayer networks were tested using the error backpropagation learning algorithm and activation function of the hyperbolic tangent type. The quality of the selected models was assessed based on graphical analysis of residual dispersion, the correlation between observed and estimated values (ryy), mean error (ME) and root mean square error (RMSE%). The most efficient neural network for predicting DBH (diameter at breast height) growth uses the DBH parameter measured at approximately two years of age and their respective means as input variables. The model developed for predicting DBH of individual trees in a specific site provided good accuracy while maintaining a correlation above 0.90 and RMSE% below 10% in most places where it was applied, without needing to adjust the synaptic weights. In conclusion, the Artificial Neural Networks with a hidden layer that use only the parameter diameter at breast height (DBH) at an age close to two years and their respective mean as input variables, is promising, but their accuracy decreases when applied to environments with climatic conditions different from those of the network training site.

Zitieren

APA

Silva, W. D. M., Araujo, M. J. D., Campoe, O. C., & Paula, R. C. D. (2026). Artificial Neural Networks for early prediction of eucalyptus stem diameter across climatic gradientes. Scientia Forestalis, 54, 1-9. https://doi.org/10.18671/scifor.v54.08

Vancouver

Silva WDM, Araujo MJD, Campoe OC, Paula RCD. Artificial Neural Networks for early prediction of eucalyptus stem diameter across climatic gradientes. Scientia Forestalis. 2026;54:1-9. doi:10.18671/scifor.v54.08

BibTeX
@article{Silva2026,
  author = {Silva, William de Medeiros and Araujo, Márcio José de and Campoe, Otavio Camargo and Paula, Rinaldo Cesar de},
  title = {Artificial Neural Networks for early prediction of eucalyptus stem diameter across climatic gradientes},
  journal = {Scientia Forestalis},
  year = {2026},
  volume = {54},
  pages = {1--9},
  doi = {10.18671/scifor.v54.08}
}
Kostenlos starten →