Flemish Forest Map 2000

Evaluation of the applicability of very high resolution imagery for updating the Flemish Forest Map 2000

Funding agency: Flemish Community, Ministry of Environment, Nature and Energy, Nature and Forest Agency

Researcher: Frieke Vancoillie

Promoter: Robert De Wulf

Duration: 10/2003 – 03/2006

Project objectives

The principal objective of this project is the development of a semi-automatic procedure to update the Flemish Forest Map 2000, based on IKONOS imagery.

The increased level of spatial detail found in VHR data has a pronounced impact on the way thematic information is extracted. As the average size of real world objects like trees exceeds the size of a VHR pixel, classification strategies necessarily need to rely on spectral, textural, as well as context-related approaches. The high spatial resolution is of considerable importance in fragmented forest landscapes where average forest patches are small and spatially scattered. In Flanders for example, 38% of the forest patches are less than 1 ha, while 85% are less than 5 ha.

Two key publications

Van Coillie, F., Verbeke, L. and De Wulf, R. “Feature Selection by Genetic Algorithms in Object-based Classification of IKONOS Imagery for Forest Mapping in Flanders, Belgium.” REMOTE SENSING OF ENVIRONMENT 110.4 (2007): 476–487.

Van Coillie, F. M. B., Verbeke, L. P. C. & De Wulf, R. R. “Wavelet-based texture measures for semi-continuous stand density estimation from very high resolution optical imagery”. Journal of Applied Remote Sensing, 5, 053560-1 (2011)