Hello all,
I am attempting to statistically quantify the degree to which two
different satellite
sensors (World View2 and IKONOS) are capable of spectrally separating several
land cover classes.
I am using a transformed divergence index to quantify the degree to which a
sensor is capable of distinguishing between the various land cover classes.
The transformed divergence index ranges from 0-2000, with above 1900 being
considered significant.
My question is what means are there for statistically comparing two
transformed divergence scores, i.e. if for Class Pairs: 15:16 Digital Globe
is ranked at 2000 (the highest statistically possible), and IKONOS for
those same two pairs is ranked 1197 (way below statistically significant) is
there any way to quantify that intra-sensor variance?
Data Summary
Separability Listing
9= Primary rainforest
15= Secondary succession
16= Agroforestry
DIGITAL GLOBE 8 Band
Separability Listing
Bands AVE MIN Class Pairs:
9:15 9:16 15:16
1 2 3 4 1999 1997 1997 2000 2000
5 6 7
8
IKONOS 4 Band
Separability Listing
Bands AVE MIN Class Pairs:
9:15 9:16 15:16
1 2 3 4 1159 677 1602 677 1197
Thanks in advance,
David
--
Doctoral Candidate
Anthropology Department
University of Georgia
