Seminar on Computational Learning and Adaptation


 
Predicting Forest Stand Height and Canopy Cover
  from LANDSAT and LIDAR Data Using Data Mining Techniques


Saso Dzeroski
Department of Knowledge Technologies
Jozef Stefan Institute
Ljubljana, Slovenia
        http://www-ai.ijs.si/SasoDzeroski/SasoDzeroski.html


The motivation for this study was to use data-mining techniques to improve the consistency, accuracy, and spatial resolution of supporting information for the forest monitoring system in Slovenia. Specifically, we aim to generate raster maps with 25m horizontal resolution of forest stand height and canopy cover for the Kras region of Slovenia.  To this end, we learn predictive models based on multi-temporal Landsat ETM+ data, calibrated by data on forest stand height and canopy cover extracted from high-resolution airborne laser scanning data. Visual inspection of the resulting maps by a forestry expert showed that they corresponded to the actual forest cover in the Kras region, both in terms of forest stand height and canopy cover.


Date: Thurs, May 25

Time: 4:15-5:30PM 

Place: Cordura Hall, Rm 100


Return to the seminar schedule