Comparison of interferometric and stereo-radargrammetric 3D metrics in mapping of forest resources
Accurate forest resources maps are needed in diverse applications ranging from the local forest management to the global climate change research. In particular, it is important to have tools to map changes in forest resources, which helps us to understand the significance of the forest biomass chang...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2015-04-01
|
| Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Online Access: | http://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XL-7-W3/425/2015/isprsarchives-XL-7-W3-425-2015.pdf |
| Summary: | Accurate forest resources maps are needed in diverse applications ranging from the local forest management to the global climate
change research. In particular, it is important to have tools to map changes in forest resources, which helps us to understand the
significance of the forest biomass changes in the global carbon cycle. In the task of mapping changes in forest resources for wide
areas, Earth Observing satellites could play the key role. In 2013, an EU/FP7-Space funded project “Advanced_SAR” was started
with the main objective to develop novel forest resources mapping methods based on the fusion of satellite based 3D measurements
and in-situ field measurements of forests. During the summer 2014, an extensive field surveying campaign was carried out in the Evo
test site, Southern Finland. Forest inventory attributes of mean tree height, basal area, mean stem diameter, stem volume, and
biomass, were determined for 91 test plots having the size of 32 by 32 meters (1024 m<sup>2</sup>). Simultaneously, a comprehensive set of
satellite and airborne data was collected. Satellite data also included a set of TanDEM-X (TDX) and TerraSAR-X (TSX) X-band
synthetic aperture radar (SAR) images, suitable for interferometric and stereo-radargrammetric processing to extract 3D elevation
data representing the forest canopy. In the present study, we compared the accuracy of TDX InSAR and TSX stereo-radargrammetric
derived 3D metrics in forest inventory attribute prediction. First, 3D data were extracted from TDX and TSX images. Then, 3D data
were processed as elevations above the ground surface (forest canopy height values) using an accurate Digital Terrain Model (DTM)
based on airborne laser scanning survey. Finally, 3D metrics were calculated from the canopy height values for each test plot and the
3D metrics were compared with the field reference data. The Random Forest method was used in the forest inventory attributes
prediction. Based on the results InSAR showed slightly better performance in forest attribute (i.e. mean tree height, basal area, mean
stem diameter, stem volume, and biomass) prediction than stereo-radargrammetry. The results were 20.1% and 28.6% in relative root
mean square error (RMSE) for biomass prediction, for TDX and TSX respectively. |
|---|---|
| ISSN: | 1682-1750 2194-9034 |