| Popis: |
Accurate and consistent measurement of carbon stocks and flows in forest ecosystems has recently gained global importance. This study aims to estimate the aboveground stand carbon (AGSC) using Landsat 8 OLI satellite image in pure Crimean pine stands and to compare the results of various modeling techniques. In this context, a total of 108 sample plots were firstly taken in a case study forest area. The AGSC of each sample area was calculated using a species-specific carbon equation developed for the case study area. The band values, vegetation indices, and texture values for each sample plot were also obtained from Landsat 8 OLI satellite image. The relationships between the AGSC and the band values, vegetation indices, and texture values were investigated by multivariate linear regression (MLR), support vector machine (SVM) and artificial neural networks (ANN) models. The results demonstrated that the ANN models with Bayesian regularization are better than the MLR and SVM models to estimate the AGSCin pure Crimean pine stands. Also, the band values showed better predictive performance in explaining the variation in AGSCthan vegetation indices and texture values. |
