tanbul 2004
> are Sparse
‘ban Areas,
Monitoring
sed on the
del by sub-
Vol. 23, n.
ent of clas-
eatures: the
t. J. Remote
:nsen, 2001.
pert system
arid to arid
m vol. 77,
Project -
R, project
ity, | 2004.
me (HABI-
Org
detection in
lassification
nm. Remote
REMOTE SENSING AS A MEANS OF ECOLOGICAL INVESTIGATION
P. Aplin
School of Geography, The University of Nottingham, University Park, Nottingham NG7 2RD, UK - paul.aplin@nottingham.ac.uk
KEY WORDS: Ecology, vegetation, habitat, monitoring, spatial resolution, QuickBird, IKONOS, Africa
ABSTRACT:
Much ecological analysis requires detailed spatial observation, traditionally conducted through field measurement. Remote sensing
has been tested extensively as a means of ecological investigation, but many such studies were limited by the relatively coarse
spatial resolution of the imagery used. The new generation of fine spatial resolution satellite sensors provides an opportunity for
detailed and accurate ecological studies, reducing the need for expensive ground survey. This paper covers two main topics. First,
the current status of the general field of ecological remote sensing is described, with particular reference to recent developments in
spaceborne data availability. Second, specific research findings related to habitat monitoring in southern Africa are presented.
Vegetation distributions are investigated at a range of spatial and temporal scales using various sources of remotely sensed data. The
vegetation information is then integrated with animal population data to further our understanding of the dynamic relationship
between the two.
1. INTRODUCTION
Ecology is 'the scientific study of the interactions between
organisms and their environment. (Begon e al. 1990, p. x).
Ecological investigation undertakes to understand, describe,
predict and control these organisms. Generally, such
investigation requires spatially explicit data, given the
fundamental need for knowledge about the location and
distribution of species (Turner e/ al. 2003). The traditional
means of collecting ecological data is through manual, field-
based observation. This approach has the benefit of generating
highly accurate measurements, but, due to its labour-intensive
nature, it is generally impractical for anything other than local
scale studies. The implications of ecological analysis, though,
extend well beyond the local scale, and there is considerable
need for, and interest in, ecological investigation at wider
spatial scales, from the ‘landscape’ (Gulinck er al. 2000) to the
entire globe (Los er al. 2002). Consequently, remote sensing
has become common in much ecological investigation,
providing the only realistic, cost-effective means of acquiring
data over large areas (Nagendra 2001, Kerr and Ostrovsky
2003).
While remote sensing has become a key mechanism for
generating ecological data, certain limitations exist regarding
the spatial detail of these data. Notably, until recently, the
spatial resolution of spaceborne remotely sensed imagery was
“far too coarse to detect most organisms’ (Turner et al. 2003, p.
306), preventing detailed ecological analysis. Specifically,
ecologists were restricted largely to 10 m and 20 m spatial
resolution imagery (panchromatic and multispectral,
respectively) from the Systeme Pour l'Observation de la Terre
(SPOT) satellite's High Resolution Visible (HRV) sensor, 30m
spatial resolution imagery from Landsat's Thematic Mapper
(TM), and other coarser spatial resolution imagery (e.g.,
Franklin et al. 1994, Luque 2000). However, given that ‘many
phenomena of interest to ecologists... occur over large extents
but at local scales’ (Read et al. 2003, p. 592), the level of
spatial detail provided by SPOT HRV or Landsat TM imagery
is likely to be insufficient for much ecological investigation.
Recently, a new generation of fine spatial resolution satellite
sensors has emerged (Van der Meer ef al. 2002, Aplin 2003a),
325
capable of providing imagery with a level of detail that may be
sufficient for meaningful and accurate local scale ecological
investigation (de Leeuw er al. 2002, Clark ef al. 2004b). In
particular, imagery with a spatial resolution of 1 m
(panchromatic) and 4 m (multispectral) or finer is available
from instruments such as IKONOS and QuickBird (Sawaya ef
al. 2003). Given that these ‘observations are at a spatial scale
equivalent to field measurements typically carried out in
ecological and land cover research’ (Goward ef al. 2003, p. 80),
the implications for ecological investigation are significant.
This paper reviews the general field of ecological remote
sensing, with particular reference to recent developments in fine
spatial resolution satellite sensors, and presents ongoing habitat
monitoring research in southern Africa. Initially, remote
sensing-based ecological investigation is introduced generally,
followed by a discussion on the ecological implications of fine
spatial resolution imagery. Next, African applications of
ecological remote sensing are described, leading finally to a
summary of ongoing habitat monitoring research in Kruger
National Park, South Africa.
2. REMOTE SENSING AND ECOLOGY
The relationship between remote sensing and ecology is not
particularly well-defined and is almost certainly under-
exploited (Gulinck ef al. 2000). A mismatch in aspirations and
practices seems to exist between the two disciplines, preventing
close integration. Ecologists, in general, seem reluctant to adopt
new approaches, particularly involving observation at, from
their viewpoint, relatively coarse spatial scales (Turner er al.
2003). Remote sensing specialists, for their part, have perhaps
focused on technological issues as their principal concern,
rather than ecological problems. Closer integration, likely
involving creativity and compromise, may benefit both
disciplines.
Despite the basic divergence between remote sensing: and
ecology, there are many successful examples of ecological
remote sensing applications. Turner ef al. (2003) describe the
distinction between direct and indirect approaches, also referred
to by Nagendra (2001). Direct ecological remote sensing
