International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
Figure 8. Graph of atmospheric absorption for low wavelengths
(from Hyvarinen, 2000)
4. PRODUCING THEMATIC MAPS
The data received from different sensors have different spatial
resolutions and it is difficult to solve problems of practical
interest only with data coming from a single band permit to
solve. Thus comes about the necessity of succeeding in
combining information supplied by different channels in order to
obtain thematic maps which allow for monitoring and evaluation
of the differing aspects of a given phenomenon. The. HSV
algorithm, described in paragraph 2, allows for “merging”
information supplied by different channels without however
altering its radiometric characteristics.
The results obtained therefore will be higher spatial resolution
images than the input data and will no longer contain single band
information but the combined data coming from three different
channels. Combining the information occurs through the use of
false colour images, as was also proposed by Karlsson (Karlsson,
1997). Each of the three basic colours (Red, Green and Blue) is
assigned a different channel.
The results of the experimentation described in this study were
obtained by combining the different channels of the MODIS
sensor (Hyvarinen, 2000), and using software ENVI ver 4.0.
Figure 9a shows the image obtained from the combination of
bands 1, 4, 3, that correspond respectively to the frequencies of
Red, Green and Blue, with which it is possible to obtain a True
Figure 9a. True Color MODIS (500m)
724
Colors image of the imaged area with nominal precision of 500
m. Figure 9b however, was obtained by using the HVS algorithm
to merge the preceding bands with band 2, in order to obtain
ground level precision equal to 250 m.
The images in figures 10a and 11a, respectively combinations of
channels 7, 4, 3 and 7, 2, 1, also have spatial precision equal to
500 m, while the images in figures 10b and 11b. which were
obtained through Image Fusion, have a spatial resolution of 250
m. Projecting the images shown in figures 10b and 11b onto
channel 2, at a higher resolution, allows for a more accurate
analysis of the spread and health of woodland areas. In fact, in
figure 10b, the lighter red areas show areas of healthy vegetation
while the areas in darker red are areas in which there is no or
diminishing amounts of vegetation. This type of representation
allows us to gain an idea of the distribution and the development
of areas in the process of desertification and also gives us more
detailed information on the typology of vegetation present.
The images obtained from the sensors mounted on the Terra and
Aqua satellites not only supply useful information for territorial
analysis but also allow for analysis and study of Earth's
atmosphere, as shown in the images in figures 12a and 13a
(obtained by combining channels 20, 31, 32 and 1, 6, 31 of
MODIS). In figure 12b it can be seen how it is possible to easily
identify the different typologies of clouds, their different altitudes
and, in the case of snow, its spread over an area of land. The data
shown in the image in figure 13b allows for an assessment of the
presence of water vapour in the atmosphere and thus facilitates
the determination of those areas at risk from fog. The same
results presented in the last figures were also obtained by
combining channels 3, 4, 5 and 1, 3, 4 of the AVHRR sensor that
correspond to channels (20, 31, 32) and (1, 6, 31) of the MODIS
sensor. In this case it was passed from a spatial resolution of
I Km (figs. 14a) to one of 250 m (figs. 1 4b).
Information can moreover be synthesized using a thematic map
that shows in a clear way areas with a presence of vegetation. Ad
example of this representation is in figure 15 where the blue
zones are areas with a large wooded presence. This result is ob-
tained automatically with the classification algorithm K-Means,
already implemented in ENVI.
Figurae 9b. True Color MODIS (250)
Internati
