While multi-
available on
> topography
‘on one hand
age of radar
ight, thereby
ing, NASA's
ctive infrared
the structure
| and spatial
instrument. It
ther spatially
vel of carbon
ever, Art. 3:5
" may, under
ommitments"
, as it can be
ed to support
ar.
0. The use of
carbon stock
ons. Archives
ould limit its
and archives
n order to be
rotocol.
90 baseline is
scale (100 m
enqvist et al.
deforestation
in greenhouse
and forestry
ges in carbon
2 furthermore
ng" between
int step is the
| and relevant
ivities, image
n, in order to
Rosenqvist, A.
minimize the effects of seasonal artefacts in the data. A spatial resolution better than the minimum area of interest - still to
be defined - will be required for this task.
assive] Optical (Multi-spectral and Panchromatic) Svstems
Optical systems are sensitive to parameters related to the structure and closure of the vegetation canopy, such as e.g.
canopy projected cover (CPC) and leaf area index (LAI), which are affected during ARD activities and fire. Detection and
spatial quantification of deforestation (D) activities, which bring about the removal of the forest canopy, is the most
straight-forward part of the three ARD components, and both panchromatic and multi-spectral remote sensing data are
deemed useful for this task. High resolution systems will be required to detect partial deforestation activities, such as
selective logging and thinning. Reforestation (R) is more difficult to detect, as it represents a gradual change from non-
forest to forest, spanning several years. Afforestation (A) events, which can be expected to take place in relatively small
patches outside the "forest" areas will be most difficult to detect of the ARD components. Multi-spectral systems are
however sensitive to growth parameters such as APAR (absorption of photosynthetically active radiation), which peaks
during the regeneration stages, thus indicating the location of potential R and A areas after the trees are large enough. High
resolution multi-spectral systems will be required for both R and A, but repetitive (annual) measurements will be essential.
Persistent cloud coverage in some areas constitutes a major obstacle. Nevertheless, the simple identification that ARD
activities have taken place can be achieved (Justice ef al. 1996) and is a valid contribution in the context of the protocol
Active fire events can be detected in an operational manner both at global (Dwyer ef al. 1998, Stroppiana et al. 1999)
and regional scale (Barbosa et al. 1999, 1998) by coarse resolution optical sensors, which provide daily coverage. Spatial
quantification of the burnt areas can thereafter be assessed with the use of high resolution sensors. The World Fire Web
network provides near-real time information on global fire activities using NOAA AVHRR (Pinnock and Grégoire, 2000).
Active Microwave Sensors.
Microwave sensors are particularly sensitive to detecting changes between images acquired at different times, even in
areas of topographic terrain, provided that the viewing geometry is kept the same. Instruments operating with long
wavelengths (L-band or longer) are more suitable for forest related monitoring than short wavelength sensors (C-band or
shorter) as the L-band signals interact with the forest at branch and trunk level, while the main interaction at C-band occurs
with the canopy. Rough soil and herbaceous vegetation may in the latter case be confused with forest.
Since SAR image acquisitions are independent of cloud cover, it is possible to accurately plan the timing of the data
takes, thus optimizing the conditions to detect change in the land cover. While it is possible to use short wavelength band
SAR systems for the detection and spatial quantification of deforestation (D) events, the use of single polarization C-band
data has proven to be problematic as forest and non-forest areas cannot always be differentiated. Still, interferometric C-
band tandem data, in particular the phase coherence, may under certain circumstances constitute a valuable source of land
cover type information (Wegmiiller ef al. 1997). If limited to single polarization data sets, the detection and quantification
of deforested (D), reforested (R) and afforested (A) areas is best addressed using longer wavelength band SAR systems,
which are more sensitive to the range of biomass associated with forests.
Polarimetric SAR systems can be expected to improve the capabilities for ARD monitoring and at least three such
systems are currently planned for the near future ALOS (L-band), Envisat (C-band) and Radarsat-2 (C-band). The
LightSAR (L-band) mission has been halted, but NASA are currently studying alternative mission concepts.
Active Optical Systems (LIDAR) :
As long as no imaging LIDAR system is available, detection of ARD activities and bum scars will be limited to those
areas sampled by the VCL. As such, the feasibility of using LIDAR to address ARD events is, as of yet, unproven.
However, LIDAR should have the ability to repeatedly characterize structural attributes at specific locations or collect
sample sets in known ARD areas which could prove useful. :
2.1.4 Quantification of above-ground vegetation biomass stocks and associated changes therein
The possibilities of making direct estimations of biomass stocks from space is naturally of prime interest in the context
of Articles 3 and 12, above.
Multi-spectral Systems. ; ; :
Direct measurements of total above ground forest biomass stocks or changes in such is not feasible with (passive)
optical systems. However, indirect estimations of biomass change is possible to a limited extent using photosynthetic
indices based on photosynthetically active radiation (PAR). PAR measurements have been routinely made using multi-
spectral sensors and may be combined with environmental data and forest growth models to predict NPP (net primary
production) which is presented in terms of units of carbon.
Active Microwave sensors. ;
The application of radar systems to measure and detect changes in above-ground biomass stocks is an active area of
research and development. There was general agreement among the workshop participants, that currently available radar
satellite systems (ERS-2 and Radarsat-1), which operate with single channel C-band, are not well suited for biomass
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1281
