Tropical peat swamp forest (PSF) has a high species diversity and value including timber value. The valuation of timber is approached by calculating the stand volume. However, the volume model at PSF is not available for every species and species groups in specific site. This study aims to obtain a volume estimation model of species groups of tree species in peat swamp forest in Central Kalimantan. The model development and validation used 120 sample trees of the dipterocarpaceae and non-dipterocarpacae species. The distribution of sample trees is 70 % for the development model and 30 % for the validation stage. Modeling used linear and non-linear models). The selection of the best model used several criteria including: coefficient of determination, relative deviation (SR <8 %) and aggregation (SA <1 %), presumptive bias (s and RSE: Root Square Errors), AIC (Akaike's Information criteriation), and NRMSE (Normalized Root Mean Square Errors). The results of this study indicated the Berkhout/diameter model as a single variable was very good in estimating the volume of trees at all levels namely species groups and all species except dipterocarpaceae groups. This model had a high coefficient of determination (R2 >95 %). The models compiled met the SA and SR criteria so that the models were very accurate in estimating tree volume at the tree level, as individuals and stands. The implication of this research was that the whole species model can be used to estimate the volume of trees in peat swamp forest. Keywords: accuracy, timber , non-linear, validation
Read full abstract