In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
■ ' 
652 
canopy or top-of-atmosphere) data sets could be generated to 
revisit the model inversion issue once again. 
3.2 Revisiting model evaluations with measurements 
RAMI was conceived as an open-access community exercise 
and will continue to pursue that direction. As such its goal is to 
move forward in a manner that addresses the needs of the 
majority of RT model developers and users. With every model 
having its own implementation of ‘reality’ it has become 
necessary to provide as detailed descriptions as possible of 
increasingly realistic canopy architectures. During the fourth 
phase of RAMI explicit 3-D tree generations (accounting for the 
position and orientation of every single leaf, twig and branch) 
were generated on the basis of detailed forest inventory data and 
L-system based or interactive tree generation software tools, 
e.g., Streit (1992) and Lintermann and Deussen (1999). Such 
tree representations, although realistic looking by design, are 
not exact copies of the trees actually present at the test sites. At 
best the RAMI-IV canopies agree in terms of the location of the 
overstorey trees and their outer dimensions and leaf content 
with what was present at the actual stands. Foliage orientation, 
distribution and colour, as well as, the branching patterns and 
densities in trees, however, are in all likelihood different. The 
same is also true for the directionality of scattering interactions 
between the sunlight and foliage, branch or background 
components, or, the directionality of the incident radiation. The 
apparent realism of some of the RAMI-IV test cases is thus at 
best an example of the capabilities of some of the 3-D canopy 
RT models but not proof of our abilities to generate structurally 
and radiatively accurate replicas of existing forest stands (that 
are suitable for the validation of canopy RT models). 
The fourth phase of RAMI has shown that 3-D canopy RT 
models are capable of representing forest stands over 1 hectare 
or more where every single leaf/needle is accounted for. The 
time thus may have come to revisit our capabilities in building 
spectrally and architecturally accurate replicas of actual forest 
sites. Earlier efforts in this direction, like the work of Martens et 
al, (1991) and those associated with large field campaigns like 
BOREAS (Sellers et al., 1997) and/or the Kalahari transect 
(Scholes et al., 2004), were not providing sufficient structural 
and spectral details to allow for an unambiguous verification of 
RT model simulations against remotely sensed observations 
over actual test sites. Structural clumping - occurring at various 
scales within the canopy - may have a significant impact on the 
BRDF of a vegetation target and thus a very fine description of 
plant architectures are needed for model verification purposes. 
Recently, Coté et al, (2009) showed that terrestrial laser scans 
could be used to generate faithfull reconstructions of individual 
trees that - when ingested into state-of-the-art 3-D Monte Carlo 
ray-tracing models - yielded accurate simulation results whether 
for in-situ observations, like those acquired by hemispherical 
photography, or for medium spatial resolution optical space- 
borne sensors. In addition, upward pointing field goniometers 
now exist that can be deployed to provide multi-spectral 
characterizations of the incident radiation field at a given test 
site at the time of satellite overpass. What is still needed 
perhaps are suitable protocols (and instruments) allowing to 
characterize the scattering directionality of foliage, wood and 
background material (as well as their spatial variability) in a 
manner that is both efficient and independent of the 
illumination conditions at the target site. 
One way top evaluate the fidelity of such ‘virtual validation 
sites’ would be to use credible canopy RT models (having 
known uncertainties) to simulate atmospherically-corrected air 
or space borne observations acquired over the same canopy 
targets under the proviso that both the characteristics of the 
remote sensor(s) and the directionality of the incident solar 
radiation at the time of overpass were accurately known. In this 
way canopy RT models could actively contribute toward the 
systematic validation of remote sensing data, products, and field 
protocols as promoted by the Committee on Earth Observation 
Satellites (CEOS). 
4. CONCLUSION 
Through a decade of systematic benchmarking efforts RAMI 
allowed to 1) identify a series of ‘credible’ canopy RT models, 
2) generated ‘community’ reference data sets, 3) establish web- 
based benchmarking facilities, and 4) increase the realism of the 
simulated plant environments. A variety of thematic domains, 
spectral regions and individual sensors could all benefit from 
being included in future RAMI activities. Due to rapid 
improvements in space sensors and physically-based retrieval 
algorithms systematic RT model validation activities are 
essential to document whether the quality of space derived 
information is improving. Here, a more proactive support from 
space agencies, scientific bodies and policy makers may help, 
for example, by making the provision of funding conditional on 
quality certificates that testify as to the aptitude of models 
and/or algorithms contained in a given proposal. Automated 
web-based benchmarking facilities, like the ROMC, can already 
now deliver such quality assurance support. 
REFERENCES 
Chen, J. M., Nilson, T. and Strahler, A., 2000, Recent Advances 
in Geometrical Optical Modelling and its Application, Remote 
Sensing Reviews, 18,227-262. 
Coté, J-F., Widlowski, J-L., Fournier, R. A., Verstraete, M. M., 
2009, The structural and radiative consistency of three- 
dimensional tree reconstructions from terrestrial Lidar, Remote 
Sensing of Environment, 113, pp. 1067-1081. 
Disney, M. I., Lewis, P. E., North, P. R. J., 2000, Monte Carlo 
Ray tRacing in Optical Canopy Reflectance Modelling, Remote 
Sensing Reviews, 18, 163-196. 
Franklin, J., and Duncan, J., 1992, Testing the Li-Strahler Four- 
component Canopy Reflectance Model in the Hapex-sahel 
Shrub Savanna Sites Using Ground Reflectance Data, In: 
Proceedings of the IEEE International Geosciences and Remote 
Sensing Symposium, pp. 200-202. 
Goel, N., 1988, Models of Vegetation Canopy Reflectance and 
their use in Estimation of biophysical parameters from 
reflectance data, Remote Sensing Reviews, 4(1), 1-222. 
Goel, N., and Kuusk, A., 1992, Evaluation of One Dimensional 
Analytical Models For Vegetation Canopies, In: Proceedings 
of the IEEE International Geosciences and Remote Sensing 
Symposium, pp. 505 - 505. 
Goel, N., and Thompson, R., 2000, A snapshot of canopy 
reflectance models and a universal model for the radiation 
regime, Remote Sensing Reviews, 18, 197-225. 
Hu, B. X., Lucht, W. and Strahler, A. H., 1999, The 
interrelationship of atmospheric correction of reflectances and