Modelling timber volume and other forest parameters using LiDAR and field data: A case study for part of Bergama State Forest Enterprise

Abstract

One of the biggest challenges in forest ecosystem planning is the measurement of forest inventories. Traditional methods of field measurements usually take time and costs lots of money. With the latest developments in remote sensing, precise estimation of some key forest parameters is becoming a reality. One of the new technologies that is being used for this is light detection and ranging (LiDAR). In this study, LiDAR derived tree height metrics as well as canopy density metrics as independent variables were regressed against the volume per ha, mean height, dominant height and number of trees per ha as dependent variables using SPSS and Excel. A total of 40 sample plots dominantly composed of Pinus brutia (Turkish red pine) were used. After processing the LIDAR data, a canopy height model (CHM) and canopy density model were obtained from which height and density metrics were derived respectively for the 40 sample plots. The best regression models obtained using LiDAR data alone had adjusted coefficient of determinations (R2) of 0.66, 0.73, 0.83 and 0.83 for volume per ha, trees per ha, average height and dominant height and RMSE of 38.39 m3 ha-1, 109 trees ha-1, 1.68 m and 1.78 m respectively. After integrating LiDAR and WorldView-3, the best adjusted R2 was 0.70 for volume per ha and RMSE of 28 m3 ha-1. All the results were significant at 0.05 and thus credible.