Modeling Stand Volume using Landsat TM Data for Fir Stands (Abies bornmuelleriana Matth.) Located in Buyukduz Planning Unit, TURKEY

Remotely sensed data in the form of satellite images have been used for decades to estimate forest parameters in support of forest management planning (Leyk et al., 2002). Since satellite data can be repeatedly acquired with reliable data quality, methods about modeling some stand attributes with dataoriginated satellite images is appropriate for obtaining information on land cover on forest areas (Wulderand and Seemann, 2003). Based on 97 sample plots, it is aiming to model relationships between stand volume and band values based on Landsat TM data for fir stands (Abies bornmuelleriana Matth.) located in Buyukduz Planning Unit, TURKEY. Multiple linear regression models were used to predict stand volumes with band values, including TM 1 - TM 5 and TM 7, originated from Landsat TM satellite image. The regression models, including different independent variables alternatives and band values, were compared with some information criteria, e.g. the adjusted coefficient of determination (R2), with Reduced Akaike’s Information Criterion (AIC), Sawa’s Bayesian Information Criteria (BIC), Schwarz Bayesian Criteria (SBC), the root mean square error (RMSE) and Mallow’s Cp, which criteria are measures of goodness of fit for regression models. These statistical analyses were performed by PROC REG and PROC RSQUARE procedures of the SAS/ETS V9 software (SAS Institute Inc, 2004). The best results for predictive performance were obtained by multiple linear regression model including TM 2 and TM 4 as independent variables. This model, statistically significant at 95% level with model parameters, explained 54.09% of the observed stand volume variability with 634.29 of AIC, 637.02 of BIC, 640.36 of SBC, 28.69 of RMSE and -0.315 of Cp. The results showed that the Landsat TM data are beneficial to estimate forest stand volume. Thus, forest managers could use remote sensing data, e.g. Landsat TM data, for predicting stand volume and for generating maps necessary for developing forest management plans