The development of better techniques for land vegetation cover and forest ecosystems monitoring is a major requirement
for local, regional and global policy and global change science. The influence of climatic variability and anthropogenic
activities on the condition of the vegetation (agricultural fields, forests, sparse) is growing up continuously. In order to
characterize current and future state of vegetation and localize zones of changes must be defined the proper criteria.
Vegetation land cover monitoring by satellite remote sensing data is one of the most important application of satellite
imagery. Vegetation reflectance has variations with sun zenith angle, view zenith angle, and terrain slope angle. To
better providing of this these effects corrections in the visible and near-infrared region of electromagnetic spectrum, was
used a three parameters model and was developed a simple physical model of vegetation reflectance, by assuming a
homogeneous and closed vegetation canopy with randomly oriented leaves. Multiple scattering theory was used to
extend the model to function for both near-infrared and visible light. This paper aims to improve the model to be used to
correct satellite imagery for bidirectional and topographic effects. Thresholding based on biophysical variables derived
from time trajectories of satellite data was applied for classifying using Landsat TM and ETM, SAR ERS-1 imagery for
Cernica forested area in the Eastern part of Bucharest town, Romania. Classification accuracies are function of the class,
comparison method and season of the year.