Department of Software Technology
Vienna University of Technology
Forest tree Mortality Simulation in Uneven-Aged Stands Using Connectionoist Networks
We describe the application of neural networks adhering to the
unsupervised learning paradigm to individual tree mortality prediction
within forest growth modeling.
The data set used for this study originates from permanent sample plots in
uneven-aged Norway spruce (Picea abies L. Karst)
stands in Austria.
In addition we use the same data set and parameterise a LOGIT model
which represents the conventional statistical approach for predicting
individual tree mortality.
Finally, we evaluate the mortality predictions using an independent data set
and compare the findings with observed mortality rates.
The encouraging results indicate that neural networks perform slightly
better than the conventional LOGIT approach.
Up
Comments: rauber@ifs.tuwien.ac.at