vegetation typing with GRASS 4.1

[Russell Yost]

We are getting very small DN values for TM raw data:
System: SunOS 4.1.3
GRASS version 4.1
Problem:
    When the bands 1-5 and 7 are displayed in the form of a
histogram,  DN values never exceed the range 1-31.  In fact,
the histogram display seems to be lumping these categories
into four greyscale levels.  It appears that a great deal of
the information contained in the data set was lost when I
imported the files into GRASS.  Furthermore,  after
examining the statistics obtained from running an
unsupervised classification in GRASS,  equally puzzling
numbers appeared.  For example, mean DN values for clas ,
representing some type of forest on the ground, never exceed
10.  Previous work with TM data sets tell me that highly
vegetated areas usually have a high mean DN value in Band 4.
This data set tells me that the mean DN for class 1 in band4
is 6.673981, and this number doesn't make sense.  In
addition, class 15, representing clouds, has the highest
DNs: low 15.983657 (band1) to high 28.832482 (band2).

    The bands were imported by first converting them from
PGA files to sunraster format in xview.  A flag in the xview
window signaled that the images "contained all 32
colors" .  When I tried to save these bands to the
greyscale option the operation failed.  In any case,
after all the initial parameters were set in GRASS, the
files were imported using the "r.in.sunrast" command.

   Any suggestions would be much appreciated.

Jonathan Deenik
c/o rsyost at Hawaii.Edu


-----
Russell Yost                                Office: 808-956-7066
University of Hawaii at Manoa               FAX:    808-956-6539
1910 East West Road                         rsyost at uhunix.uhcc.Hawaii.Edu
Honolulu, HI, 96822