MONITORING BIODIVERSITY AND LANDSCAPE RICHNESS 
WITH DIGITAL EARTH IMAGERY 
Richard H. Podolsky 
The Island Institute, 60 Ocean Street, Rockland, Maine, USA 04841 
ABSTRACT 
Measuring and monitoring biodiversity and landscape richness for large 
portions of the earth is becoming increasingly important due to the acceleration 
of human impact on ecosystems. However, methods to do so have eluded 
ecologists. Here I present a method to estimate landscape diversity for large 
sections of the earth directly from digital imagery. 
Landscape diversity was calculated by applying the Shannon-Weaver diversity 
index to data extracted from digital earth imagery. This index measures 
diversity by examining the “predictability” in a given data stream. The data 
supplied by the digital imagery to this diversity model are the area 
measurements for each spectral class in all or a portion of a SPOT multispectral 
image. Spectral classes appear to correlate well to specific habitats. 
Thus, the diversity of spectral classes for any point on earth is assumed to be a 
measure of habitat or landscape diversity. Area measurements for each habitat 
are extracted directly from the SPOT imagery via a pixel tallying routine 
mediated by GAIA software on a Macintosh II computer. 
Landscape diversity and richness was calculated for the 423 islands in a SPOT 
Multi Spectral (MS) image. Generally, patterns of island diversity yielded 
results consistent with island biogeography theory. For example, landscape 
diversity of islands correlated positively to island size and convolutedness of 
island shorelines. Data also yielded a positive correlation between island 
landscape diversity and the richness of mammal species on 18 islands. 
Monitoring and ultimately ensuring the biodiversity of the earth is of critical 
importance and it appears that digital earth image data sets and other remotely 
sensed data can play a vital role in this endeavor. 
KEY WORDS: diversity, biodiversity, landscape, island, digital, imagery, global- 
monitoring. 
INTRODUCTION 
The richness and diversity of 
ecological systems have long 
fascinated ecologists. This 
fascination has focused on both the 
theoretical, for example, the 
relationship between a given 
system’s diversity and its stability, 
and the practical implications of 
diversity. On the practical side, the 
general perception of ecologists and 
conservation biologists is that, all 
things being equal, diverse systems 
are more worthy of preservation 
relative to simpler systems (Wilson 
1988). The reason for this is that 
diverse ecosystems appear to 
support richer assemblages of plants 
and animals than do simple ones. 
However, it must be pointed out that 
many “simple” systems support rare 
or endangered species whose 
protection is also critical. 
The rate of human impact on the 
earth’s surface has greatly 
accelerated in the second half of this 
century and is likely to accelerate 
further in the next 50 years. Thus we 
are faced with having to apply a 
triage approach to the question of 
which landscapes we should be 
preserving. Consequently, a 
methodology that quickly identifies 
regions of the earth with high 
diversity is of keen interest. The data 
and methodology reported here was 
motivated by this need for an efficient 
way of extracting estimates of 
relative diversity or richness at the 
landscape level. 
Ecologists recognize three levels of 
diversity: alpha diversity (also called 
species diversity) which measures 
taxon diversity within a given 
ecosystem or habitat; beta diversity 
(also called landscape diversity) 
which measures diversity of habitats 
or landscapes within an ecosystem; 
and gamma diversity, which is the 
42