|Title||Radar remote sensing to support tropical forest management|
|Author(s)||Sanden, J.J. van der|
|Source||Agricultural University. Promotor(en): R.A. Feddes; R.A.A. Oldeman; D.H. Hoekman. - S.l. : S.n. - ISBN 9789054857785 - 330|
Biological Farming Systems
|Publication type||Dissertation, internally prepared|
|Keyword(s)||bosbouw - remote sensing - luchtkarteringen - tropen - bosbedrijfsvoering - planning - tropische regenbossen - vegetatie - toepassingen - forestry - remote sensing - aerial surveys - tropics - forest management - planning - tropical rain forests - vegetation - applications|
This text describes an investigation into the potential of radar remote sensing for application to tropical forest management. The information content of various radar images is compared and assessed with regard to the information requirements of parties involved in tropical forest management at the global, national and local spatial levels. The study distinguishes between the use of radar remote sensing for application to forest resource assessment and forest resource monitoring. Both assessment and monitoring are essential components of procedures for sustainable forest management. The radar data studied are of tropical forest areas near the township of Mabura Hill in Guyana and the city of San José del Guaviare in Colombia. Mabura Hill is comprised of differing intact, primary forest types and forests that have been subjected to industrial selective logging. San José del Guaviare, on the other hand, is characterised by the presence of secondary forests and a variety of non-forest cover types. The available radar data set includes high resolution airborne radar images with differing wavelengths (i.e. X-, C-, L- and P-band) and polarizations, time-series images acquired by the first European remote sensing satellite ERS-1 and a collection of low altitude, nadir-looking, X-band scatterometer measurements.
The study makes use of three fundamentally different information sources from the radar return signal: its strength or backscatter, polarization and phase, and spatial variability or texture. Results show that backscatter values computed from L- and P-band radar data and textural attributes computed from high resolution X- and C-band radar data make modest to good and complementary bases for region-based classification of tropical land cover at the level of primary forest types. Textural attributes and backscatter values computed per region from mono-temporal ERS-1 images make modest bases for classifying at the levels of primary forest, logged-over forest, secondary forest and non- forest and poor bases for classifying at the level of primary forest types. Roads are usually the most easily observable indicators of foregoing and/or forthcoming (selective) logging and other human activities in ERS-1 images. Detection of change in road networks by means of ERS-1 images would make a good first step in forest resource monitoring at the national spatial level, in particular. Textural attributes enable the ranking of forest types according to the degree of canopy roughness. Specific textural attributes also allow for quantification of canopy architectural properties. Despite differences in measurement scale, the canopy roughness of the land cover types studied was found to appear similarly in the texture of the available spaceborne and short wavelength airborne radar images.