RS, Vol. XXXVIII, Part 7B 
In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
jectra for a constructed 
tails. 
ion did a good job on the 
t as good for the offset, 
d the simulation, had the 
ry little of the spectral 
trum lower than the hole 
feature depth. The broad 
ell simulated, impact of 
;tors. 
2.4 Roughness Retrieval from Directional Views 
For practical applications, correcting for roughness effects on 
retrieved temperatures and emissivities requires remotely 
estimating surface roughness at sub-pixel scales. One approach 
that has been tested (Mushkin & Gillespie, 2005) uses bi 
directional VNIR imaging, such as is available from ASTER, to 
estimate sub-pixel roughness at scales up to 15 m, the resolution 
of the acquired images. The approach makes use of differential 
sub-pixel shadowing in the 'down' and 'up' sun images as a 
relative proxy for roughness.. The relative measure of 
roughness is the DN ratio between the two images, corrected for 
path radiance using “dark-object subtraction”), with ratio values 
diverging from unity with increasing surface roughness. This 
ratio proxy roughness for roughness is largely insensitive to 
atmospheric effects, but must be calibrated to a quantitative 
measure of roughness, such as rms elevation. Calibration of the 
ratios to absolute values has been done from field measurement 
of micro-topography and modelling of shadows. The 
calibration is sensitive to regional topographic slope (within 5- 
10°), and sun elevation angle, and therefore requires re 
calibration for each new application. A result of the calibration 
is shown in Figure 8. Older, smoother fans are darker (less 
shadowed); parts of the dry lake are smooth salt flats, and 
others are rough pinnacles of salt ~40 cm high. 
Figure 8. This is a sub-pixel roughness image calculated 
from two ASTER images of the Trail Canyon alluvial fan in 
Death Valley National Park. Image ratios were calibrated 
to roughness using field data. 
3. SUMMARY AND CONCLUSION 
scales did not clearly show expected trends of retrieved 
emissivity spectra as a function of roughness (size), and part of 
the reason seems to be the ability of individual gravel pieces to 
maintain a temperature gradient resulting from differential solar 
heating. Modelling and model validation measurements at the 1 
to 10 cm scales show predictable changes of emissivity spectra 
with surface roughness: emissivity goes toward a black body for 
rough surfaces and in comers and shapes with a strong three 
dimensional form. The model is an abstraction, and its heat 
diffusion model is simplified. For many surfaces, this has not 
been a problem, but for surfaces with complex geometry (more 
than one value of z for an x, y location) or where three- 
dimensional heat diffusion is important, simulations of mean 
temperatures break down. Radiosity-produced variations can 
still be simulated. Although not discussed here in detail, 
compensation for roughness effects is possible given two or 
more images of the same area from different positions (with 
about the same resolution) and given knowledge of the 
roughness or valid simulations. 
Multiple radiative interactions between surface elements do 
tend to drive observed spectra toward a blackbody spectrum 
even though the material properties are constant. The impacts 
are significant but variable and usually don’t overwhelm the 
signal. The effects need to be quantitatively understood in order 
to understand thermal spectral measurements of most surfaces 
in the environment. Flowever, surface roughness, while 
important, is one factor among many that modulate both the 
magnitude and spectra of ground-leaving thermal radiance and 
needs to be considered in context. 
REFERENCES 
Balick, L. K., M. E. Howard, H. M. Gledhill, A. Klawitter, and 
A. R. Gillespie, 2009. “Variation and sensitivity in spectral 
thermal IR emissivity measurements,” IEEE WHISPERS, 
Grenoble, France. August 26-28, 2009. 
A2-Technology, 2010, 
http://www.a2technologies.com/exoscan_handheld.html 
Nanovea, 2010, http://www.nanovea.com/Profilometers.html 
Mushkin, A. & Gillespie A. R. (2005). Estimating sub-pixel 
surface roughness using remotely sensed stereoscopic data. 
Remote Sensing of Environment, 99 (1-2), p.75-83 
Salvaggio, C., and C. J. Miller, 2001, “Methodologies and 
protocols for the collection of midwave and longwave infrared 
emissivity spectra using a portable field spectrometer,” SPIE, 
Image Exploitation and Target Recognition, Algorithms for 
Multispectral, Hyperspectral, and Ultraspectral Imagery VII, 
Volume 4381, April 2001. 
Four studies look at the effects of surface roughness on the 
energy emitted by that surface. These studies cover spatial 
scales from sub-millimetre to tens of meters, and the effect of 
roughness across these scales is due to radiative interactions 
between surface elements. (There can also be a temperature 
effect, not discussed here.) The four studies represent different 
approaches to understanding the effects of surface roughness on 
thermal IR emittance. At sub-millimetre scales, roughness 
changes from sanding rocks alter diffuse reflectance by nearly a 
factor of two across the spectrum for the surfaces studied. 
Precise field measurements of radiance of gravel at centimetre 
Telops, 2010. 
http://www.telops. com/index. php?option=com_content&view= 
article&id=60&Itemid=59&lang=en 
ACKNOWLEDGEMENT 
This work was funded by the U. S. National Nuclear Security 
Administration, Office of Nonproliferation Technology 
Development and Treaty Verification, under contract DE- 
AC52-06NA25396 with Los Alamos National Security, LLC. 
LA-UR 10-01283