Spatial Prediction of Some Biological Forest Variables by Terrain Analysis –the Kheiroud- Kenar Forest, North of Iran

This research conducted to evaluate spatial pattern of some biological variables (mean diameter, density, richness index and biodiversity index) of forest by linear regression models using terrain analysis in the Kheiroud-Kenar forest located at the Nowshahr, north of Iran. To obtain these parameters, sampling was performed on a systematic grid (250× 400 m) in 193 plots with one ha area. Density and mean diameter were computed at each plot. The richness index in each plot was estimated by extraction of species number. The biodiversity of each plot was calculated using Simpson index. The primary topographic attributes (slope, elevation, aspect, profile curvature and plan curvature) and compound or secondary attributes (wetness index, stream power index, solar radiation and LS factor) were estimated by DEM model. The multiple linear regression models by stepwise method were fitted between topographic attributes and biological variables of forest. The developed models were validated using some additional data (60 plots); and mean error (ME) and root mean square error (RMSE) were calculated to verify unbiased and accurate predictions. The result of this research indicated that richness index show a significant (r= 0.81, p<0.05) relationship with elevation, profile curvature, LS factor and aspect. The biodiversity index in the examined forest showed a significant (r= 0.71, p<0.05) with elevation, LS factor, slope and stream power index. The mean diameter and density also revealed the significant relationships with topographic attributes. The results showed that forest variables in the given area could be predicted about 50-60 % of variation by these linear models. These models can be used to predict spatial pattern for adjacent forest with similar conditions (such as management and geology) properties using DEM models and without no measurement or sampling.