Height of forest vegetation and forest density

Project was accomplished together with Geological Survey of Slovenia, Slovenian Forestry Institute, Institute Jožef Stefan, and Faculty of Natural Sciences and Engineering. It was sposored in the framwork of Target research programme »Science for Peace and Security 2006–2010«.

The main idea was based on existing studies exposing high correlation between different satellite data and vegetation height. In the project we proposed relatively accurate and inexpensive method based on remote sensing techniques for obatining spatialy explicit big-range information about two forest parameters: height of forest vegetation and forest density (canopy cover). The method assumes that primarily model is calibrated on limited area using high quality expensive reference data, to be further on unilized on full area using inexpensive input data. The method was tested on a study area of 86 km2 around town of Cerknica, which is representative of the typical large forest areas in Slovenia.

As a source of accurate reference data the lidar data was used. Model was calibrated by implementation of different machine learning algorithms. The input data for the final model-based map computation were the multispectral SPOT, IRS and Landsat satellite data as well as aerial IR imagery. As final result we proposed optimal combination of tested algorithms and imagery.