Tree Density Estimation of Forests by Terrain Analysis and Artificial Neural Network

The interaction between landscape and forest properties has been well documented, and thus it is plausible to assume that landscape factors in a forest region have a determinant function in forest properties formation. Although previous researches have identified the significant relationships between forest attributes and environmental factors, but there is no definite model available for this properties delineation. Utilizing digital terrain models and its extractable information can help for this purpose. This study was conducted to evaluate artificial neural network ability for prediction spatial distribution of forest tree density and for production continuous map using primary and secondary topographic attribute in Shastkolate forestry plan, district I. The primary and secondary topographic attributes calculated from digital elevation model with 10m resolution. Geometric coordination of plots which were recorded by GPS, mapped in GIS. Then Primary and secondary topographic attribute derived in this plot location. Tree density obtained by counting number of tree in 1 Are circle plots and then calculation in hectare. The relationship between forest tree density and terrain attributes analyzed applying two types of artificial neural network (MLP and RBF). Results showed that RBF neural network provides more accurate results than MLP neural network. Moreover, the regression analysis was done for comparison of the results of ANN models with linear models. The results verified the ability of artificial neural network for prediction of forest tree density and also indicated that ANN approach can predict approximately 65% of the forest tree density variation in the given study area using topographic attribute.