Summary:
Published in For. Chron. 96(02): 141-150.
Forest growth models are essential for forest management and the elaboration of forest policies. For model users, assessing the performance and the reliability of a model is essential. Hence, validation is an important step of model development. However, species-specific and observation year-related patterns in model error are rarely reported, although the ongoing changes in growth conditions are likely to increase the presence of such patterns. In this study, we analyzed the basal area prediction error of Artemis, a single-tree empirical growth model. Even if the model’s basal area predictions were unbiased in most conditions, we detected species- and observation year-related patterns in prediction errors. These trends were strongest for sugar maple (Acer saccharum Marsh.), for which a shift occurred from underestimation in 1975 to overestimation in 2010, and for balsam fir (Abies balsamea (L.) Mill.), for which a shift occurred from overestimation to underestimation over the same period. A better consideration by the model of soil, climate and pest disturbances could contribute to reducing model bias. These results are relevant for both developers and users, who should be aware that predictions for these species are likely to be increasingly biased as the length of the projection period increases.
Permanent identifier (DOI):
Sector(s):
Forests
Catégorie(s):
Scientific Article
Theme(s):
Forest Growth and Yield Modelling, Forestry Research, Forests
Departmental author(s):
Author(s):
POWER Hugues and Isabelle AUGER
Year of publication:
2020
Format:
PDF available upon request
How to get the publication:
Keyword(s):
Acer saccharum, Abies balsamea, Fagus grandifolia, forest growth and yield modelling, forest growth model, validation process, sugar maple, balsam fir, american beech, forestry research scientific article