Modeling some stand parameters using different remote sensing data in pure scots pine stands in Sinop Regional Directorate of Forestry

In this study, the relationships between some stand parameters (stand mean diameter, stand dominant height, number of trees, stand basal area and stand volume), leaf area index, aboveground biomass and carbon values and, the remote sensing data obtained from active (Sentinel-1) and passive (Sentinel-2, Landsat 8 OLI and Drone) sensors were modeled by multiple regression analysis (MLR) and artificial neural networks (ANN) techniques in pure Scots pine stands in Sinop Regional Directorate of Forestry. A total of 184 sample plots were taken in the study. Classical forest inventory measurements were made in each sample plot, and some stand parameters, aboveground biomass and carbon values were calculated for each sample plot. In addition, the backscattering and band values for the Sentinel-1 and the reflectance, vegetation indices and texture values for Sentinel-2, Landsat 8 OLI and Drone satellite images were calculated for each sample plot. Ground and remote sensing data of the sample plots were modeled with MLR and ANN techniques. When the modeling results obtained are examined, the best model results using textural features of Sentinel-2 15x15 window size for basal area with ANN technique (Model R2=0.84, Test R2=0.77), using textural features of Landsat 8 OLI 135° for stand volume with ANN technique (Model R2=0.83, Test R2=0.77), using textural features of Sentinel-2 7x7 window size for stand mean diameter with MLR (Model R2=0.83, Test R2=0.81), using textural features of Sentinel-2 15x15 window size for number of trees with ANN (Model R2=0.81, Test R2=0.69), using textural features of Sentinel-2 5x5 window size for stand dominant height and leaf area index with ANN (Model R2=0.71, Test R2=0.68) and (Model R2=0.74, Test R2=0.32), using textural features of Sentinel-2 9x9 window size for aboveground biomass with ANN technique (Model R2=0.82, Test R2=0.75) and using textural features of Sentinel-2 15x15 window size for aboveground carbon with ANN technique (Model R2=0.89, Test R2=0.77) were found. Also, with vegetation indices values obtained from drone images and the most successful model were found for stand basal area (R2=0.40) and stand dominant height (R2=0.44). In addition to, with the reflectance values obtained from the drone images, the most successful model (R2=0.32) was found for the stand basal area.