Modelling litter layer mass using satellite data andenvironmental variables in brutian pine forest stands

Litter layer on the forest floor has an important share in the amount of stored carbon (C) in forest ecosystems. A considerably amount of litter mass was accumulated in the conifer forests in the Mediterranean Region of Turkey. Litter mass should be regarded to estimate forest carbon with low error. Rapid and cost-effective methods should be needed for Litter mass inventory. Satellite remote sensing data covering broad geographic regions are alternative data source in this scope. Besides, environmental factors have effect on litter mass. Accordingly, it is important to reveal the relationships between litter mass and the variables derived from satellite data and environmental variables. In this thesis, we firstly investigated that the relationships between litter mass and the Normalized Difference Vegetation Index (NDVI) images derived from RapidEye, SPOT-5 and Aster satellite data and the environmental variables including temperature index (TI), radiation Index (RI) and aspect suitability Index (ASI). Statistical assessments showed that the Pearson correlation coefficients for the linear relationships between litter mass and NDVIRapidEye, NDVISPOT, NDVIAster ve TI were 0.75, 0.69, 0.73 and -0.41, respectively. Then, litter mass was estimated by means of generalized additive model (GAM) using the explanatory variables including NDVI images and environmental data. According to the cross validation test, the best model using NDVIAster and TI as explanatory variables explained a total variation of 49 percent of the response variable (litter mass). We may conclude that the litter mass on the floor of brutian pine stands can be moderately predicted using satellite data and environmental variables. We consider that the results of this study may contribute to estimation of C storages more accurately in the brutian pine stands.