Regeneration assessment in uneven-aged mixed species stands using neural networks

Estimating regeneration establishment is hampered by the difficulty in coll ecting regeneration data and random impacts in the occurrence of regenerati on. Artificial neural networks represent a computational methodology widely used to uncover the structure of a large val icr) of data. in general, one may recommend the application of neural networks in areas characterized by noise, poorly under stood intrinsic structure, and changing characteristic s. Each of those characteristics is present in predicting regeneration esta blishment within uneven aged mixed species stands. In this paper ne describ e a design and estimation procedure to predict regeneration establishment u sing data, from the experimental forest, University of Agriculture in Vienn a, Austria. The result of the study is that the number of juvenile trees pe r unit area, the relative percentage of individuals by tree species and the mean regeneration height can be predicted with neural networks. The predic tion results are more accurate than the results from the conventional stati stical approach based on regression analyses.