FOREST APPLICATIONS
SESSION
Morning, 4 April 2001
Chair : Thuy Le Toan
Co-chair : Shaun Quegan
Abstract Number : CEOS-SAR01-042
Preferences : Oral
Session : Forest Applications
Evaluation of JERS-1 SAR mosaics for hydrological applications in the Congo river basin
Ake Rosenqvist (1) and Charon M. Birkett (2)
(1) NASDA/EORC
Roppongi First Bldg. 14F, 1-9-9, Roppongi, Minato, Tokyo 106-0032, Japan, Email : ake.rosenqvist@nasda.go.jp
(2) NASA/GSFC
Mailcode 971, 20771, Greenbelt, MD, USA, Email : cmb@nemo.gsfc.nasa.gov
Two JERS-1 Synthetic Aperture Radar mosaics covering central Africa were investigated with respect their
potential usefulness for studies of seasonal flooding dynamics in the Congo river basin. The mosaic quality was
initially evaluated with respect to geometric and radiometric accuracy, as well as to the temporal composition of
the data used in the mosaic assembly. The investigation revealed no major defects that should prevent utilization
of the data for hydrological applications. The state of flooding captured in each of the two mosaics - preferably
portraying the Congo and its tributaries during their annual maximum and minimum water marks - was
furthermore evaluated by examining river stage information. Stage data contemporary with the satellite
acquisitions (1996) were derived from the TOPEX/POSEIDON Radar Altimeter and supplemented by historical
in-situ records. The results suggest that the SAR mosaics may serve well to appraise the maximum extents of
flooding in the Congo river basin, but quite poorly to assess the dynamics and ranges of the variations. It was
reckoned that a minimum of three acquisitions within one year would be required to better describe the complex
hydrology in the region. The validation uncertainty in rivers where only historical in-situ data were available was
palpable, thus stressing the utility and future potential of radar altimetry to provide substitute measurements of
river stages.
Keywords: JERS-1 SAR, GRFM, flooding
Abstract Number : CEOS-SAR01-051
Preferences : Oral
Session : Forest Applications
Contribution of SAR Systems to the Estimates of Carbon Budget in Forests
Thuy Le Toan
Centre d'Etudes Spatiales de la Biosphere (CESBIO), CNES-CNRS-Universite Paul Sabatier
18 Av. E Belin, Bpi 2801, 31401, Toulouse Cedex 4, France, Email : Thuy.Letoan@cesbio.cnes.fr
The main open questions concerning the global carbon cycle relate to the lack of knowledge of exchange
processes that link the atmosphere and the terrestrial ecosystems. To quantify carbon exchange processes,
information is required on the spatial distribution of the carbon stock and carbon fluxes from the ecosystems.
Forest ecosystems, which contain a large part of the carbon stored on land in biomass, play a major role in the
terrestrial carbon budget. Present methods for measuring and monitoring carbon stocks produce estimates of
unknown reliability. At a regional scale, methods are based upon measurements at specific locations (e.g.
experimental plots). At the global scale estimates of stocks are based on global inventories and surveys (e.g.
forest inventories published every 10 years by FAO). Net Primary Productivity (NPP) is, however, simulated and
extended both spatially and temporally by process models, but the models suffer from unrealistic input variables
and a lack of data for validation.
Satellite missions which can provide biomass information, with integrated modelling, will be a unique technique
to quantify vegetation carbon stocks and their changes, both at regular intervals and in a global and consistent
manner.
This paper will examine the relevance of biomass information the SAR systems can provide for the range of
biomass of temperate, boreal, and tropical forests, with respect to the requirements in carbon budget monitoring.
The physical background of the biomass retrieval methods will be exposed in terms of the relationships between
the biomass of tree and canopy elements and the SAR measurements (intensity, polarimetry, interferometry) at
different frequencies( C-, L-, P-, VHF-bands) and polarisations. Illustrations of biomass retrieval with the
existing spaceborne SAR data (ERS, ERS tandem InSAR, JERS), and airborne data (AIRSAR, ESAR, Carabas)
will be presented.
Observation strategy to meet the science requirements for the next decade will be discussed.
Abstract Number : CEOS-SAR01-073
Preferences : Oral
Session : Forest Applications
Detecting Fire Scars Using JERS, ERS and Radarsat
Jon Ranson (1), Guoqing Sun (2), Vyacheslav Kharuk (3), and Katalin Kovacs (4)
(1) GSFC/NASA
Code 923 Bldg. 33 G417, 20771, Greenbelt, USA, Email : jon@taiga.gsfc.nasa.gov
(2) Department of Geography, University of Maryland
Code 923 Bldg. 33 G417, 20771, Greenbelt, USA, Email (2) : guoqing@aspen.gsfc.nasa.gov
(3) Sukachev Institue of Forest
Krasnoyarsk, Russia, Email : kharuk@forest.akadem.ru
(4) Science Systems and Applications, Inc.
Email : kkovacs@forest.gsfc.nasa.gov
As part of the Siberian Mapping project, this study evaluated to what extent three different radar sensors are
capable of detecting fire scars in the boreal forest. These sensors were JERS (LHH; March 31, 1997), ERS1
(CVV; June 7, 1998) and Radarsat ScanSAR (CHH; Sept. 7, 1999). The images were geocoded and co-registered.
The location of the fire scars and the different burn classes within the fire scar were determined based on field
surveys and two Landsat 7 images taken at different dates (on July 31, 1999 and October 13, 1999).
Using Landsat 7 data the following classes could be differentiated: forest, shrub land, grasses, burned forest,
burned shrub land and burned grasses. Based on the JERS image it was not possible to distinguish burned and
unburned forests and burned and unburned grasslands. Burned shrub lands, however, appeared brighter than
unburned shrub lands. On the ERS image, all burned classes appeared bright, while all unburned classes appeared
darker. Finally, based on the Radarsat image it was not possible to distinguish any of the classes from each other.
A sensor artifact called 'scalloping' was present on the Radarsat image which might have contributed to this.
This study found that ERS data were most suited to distinguish fire scars from unburned vegetation but it was not
possible to detect the internal patterns of fire scars with this sensor. On the contrary, JERS data were best suited
to detect landscape patterns within fire scars, however, they could not be used to distinguish unburned vegetation
classes from fire scars. Radarsat data were found to be unsuitable for fire scar detection.
Abstract Number : CEOS-SAR01-074
Preferences : Oral
Session : Forest Applications
Effects of Slope and Aspect on Radar Backscatter of Forest Canopies
Jon Ranson (1) and Guoqing Sun (2)
(1) Department of Geography, University of Maryland
Code 923 Bldg. 33 G417, 20771, Greenbelt, USA, Email : guoqing@aspen.gsfc.nasa.gov
(2) GSFC/NASA
Code 923 Bldg. 33 G417, 20771, Greenbelt, USA, Email : jon@taiga.gsfc.nasa.gov
Estimation of forest biomass using SAR data is complicated in the presence of a strong topographic gradient,
because the topography influences the radar backscatter. The terrain effects on radar backscattering can be
complex. For example, changes in radar incidence angle caused by terrain slope can have several effects on radar
image data. Because of terrain the forest spatial structure changes. When trees of a relatively uniform stand grow
on a slope, the portion of the side of these trees facing the radar will be exposed to the radar. Terrain-effect
correction techniques are designed to reduce effects of incidence angle and illuminated target area. In order to
investigate the terrain effects on radar signature, a 3D radar backscatter model was modified to model backscatter
from forest stands on slopes.
The modeling results were compared with SAR data. A model-based slope correction was applied to radar
backscattering image, and the biomass estimation from SAR data in mountainous area was evaluated.