2D/3D DATA DERIVED FROM AERIAL STEREOSCOPIC IMAGERY IN THE TROPICS

It is known that tropical rainforests are the worlds largest carbon storage. However, the amount of biomass that actually exists can only be roughly estimated, especially because terrestrial forest inventory is very demanding, cost and time expensive. Forest inventory is also the only method to map biomass directly at small-scale area. High-resolution aerial imagery and LiDAR are two methods to map biomass indirectly at a small-scale level. In this project both high-resolution aerial images and terrestrial forest inventory data available were analysed for preselected areas of Suriname. The thesis aimed in linking terrestrially measured biomass data of High Dryland forest to features derived from aerial imagery including spectral information and height. In the thesis the term biomass is always equated with aboveground phyto-biomass, i.e. vegetation. The thesis opens with a short introduction and a review of the state of the art containing current biomass estimation and methods used for deviation of features, sorted according to data type. It provides a description of used methods and how they are used to derive digital surface models from stereoscopic imagery, and features such as volume, forest canopy density, number of trees and statistical features based on height. This chapter also describes the outcome of the feature derivation and the arising problems and limitations. The next part shows the results of correlation and implementation and result of regression analysis, and in conclusion, a discussion and an outlook into the future are provided.