International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
The standard method for specifying the color of soil is based on The channel emissivity difference and mean channel emissivity 
a comparison of soil samples of color chips contained in a can be calculated directly from AVHRR/NOAA data using 
Munsell color chart (Munsell, 1994). The Munsell color NDVI Threshold Method - NDVI!"M (Sobrino and Raissouni, 
designation makes use of a characterization scheme that 2000). 
describes color in terms of three variables: hue, value and 
chroma. The hue notation of a color indicates its relation to red, Vegetation indices and surface emissivity can be considered as 
yellow, green, blue and purple, according to Munsell (Munsell, a function of the ecosystem investigated, climate, terrain, soil 
1907) hue is “the quality by which we distinguish one color and hydrology variables. Conceptually the vegetation indices 
from another”. Value is a neutral axis that refers to the gray and emissivity can be modeled using those environmental 
level of the color, it is “the quality by which we distinguish a variables: 
light color from a dark one”. Chroma is the quality that 
| distinguishes the difference from a pure hue to a gray shade. 
| V] / Emissivity = f(CI, Ve, Ph, S) + K (5) 
3. THEORETICAL MODEL 
  
The sub-models may in turn be represented as a function of 
The proposed approach aims at assessing soil color from their major components: climate (C7), Vegetation/Ecosystem 
NOAA/AVHRR data. It starts from establishing the correlation (Ve), Physiography (PA), Soil/Hydrology (S). Where K, is the 
erosive models between soil color, collected in situ by pedologists, and ^ modeling errors caused by environmental variables and 
angein | Vegetation Indices and Emissivity, calculated from the NOAA potential inaccurate measurements. The model could evidently 
r from images. The next step is the inversion of the models so that soil be more complex, however, not all environmental variables are 
ibed in color can be determined directly from the NOAA data. completely independent, what makes it possible to obtain 
Upper theoretical VI/Emissivity with a limited number of 
causing It is a semi-empirical approach. The determination of the environmental variables. 
razilian correlation models between vegetation indices and emissivity 
1 PAVI and soil color/moisture starts from the definition of the physical Vegetation indexes are influenced by variations of vegetation 
digital | significance of the vegetation indices and emissivity. and soil. They can be considered as the sum of two 
| components: 
Vegetation Indices have been used extensively for the 
derivation of the biophysical properties of vegetation and soil. 
nitoring In this work a few types of vegetation indices were used, in 
order to determine witch is more useful for the assessment of 
color: 
Vr VES ¥ Tvecetation (6) 
The same can be said about surface emissivity: 
* Normalized Adjusted Vegetation Index (NDVI); 
air and NDVIE= (Die + Prin) i * Died) l eb. , 
fies I (Pai Pred (Pair Pred ( ) 8.7% £soil + Evegetation (7) 
of the 
teristics Red Soi sie 3getati sx J Netz 
; : Ned Seit Adiusted Vegetation Index (MSA VI) (Oi et ai; In this work we segmented the images and investigated only the 
M : locations where the influence of soil in the indices are greater 
| : 2 0.5 
i = ir + - TR + - ir - 2 2 I : * : : 2 > 
rminant MSAVI = (@pnir + 1) - (@pPnir + 1° - 8(Pnir = Prea)) 7) / (2) than that of the vegetation. So, we restricted the application of 
on wilh the model to the space of vegetation indices where the influence 
fact it is of the component VI,egetation 1S SMAll (NDVI between 0 e 0.2). 
à * Global Environment Monitoring Index (GEMI) (Pinty and 
ilar soi 
Orefraoip 3. 5 = ae : s 
Verstraete, 1992); Furthermore, for a specific geographic location, the 
  
DEM] A ; 92st) : (pea = 0.125) 741 - Bra) 3) vegetation/ecosystem and phisiography sub models become 
e f - z 3 3 . . . ~ ~ (9 . 
e of the E=(2(Pnir - Pred’) + 1.5Pnir + 0.5Prea) / (Pnir + Prea + 0.5) relatively less time variant. Therefore, NDVI for a specific time 
ppraisal. | (f) at a specific geographic location becomes primarily a 
ned on | function of climate variables and soil moisture/color. 
et ai e Purified Adjusted Vegetation Index (PAVI) (Singh et al., 
«cdi 9 AS 
re and 2009). ; > ; . tong ; doi i 
nina el PAVI = (Pair. - Pred ) /(Prir. = Pred’) (4) VI(t) = f [soil color/moisture] + f [climate variables] + K1 (8) 
1999). 
ed color] Surface Emissivity is a measure of the inherent efficiency of the To simplify the models: 
calcium | surface in converting heat energy into radiant energy above the 
surface. It depends upon the composition, roughness and 
moisture content of the surface and on the observation VI = f [soil color] + f [temperature)] + K2 (9) 
conditions (i.e. wavelength, pixel resolution and observation 
as 8 angle). Surface emissivity variation, consequently, have a direct 
and the relationship with surface composition change (Sobrino et al., Emissivity = f [soil color] + K3 (10) 
[vers are 2000). | 
different Surface temperature was calculated through one of the split 
window algorithms (Ulivieri et al, 1992); 
wers the 
221