In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
TROPICAL BIODIVERSITY MAPPING FROM
HYPERION IMAGE IN BOGOR INDONESIA
A. Wijanarto* 3 , F. Amhar 3
a Geomatics Research Division, National Coordinating Agency for Surveys and
Mapping , Jl. Jakarta Bogor km 46 Cibinong Bogor, 16911, Cibinong, Indonesia
wij anarto_ab @ yahoo .com. au, famhar @ yahoo .com
Technical Commission VII Symposium 2010
KEY WORDS: Hyper spectral, Land Cover, Mapping, Soil, Vegetation
ABSTRACT:
Hyperspectral remote sensing is increasingly used for many applications, and one of them is biodiversity mapping, common used
data are MODIS and other low resolution satellite images, which have been freely accessed and obtained, while the use of medium
resolution hyperspectral images was rare, until mid 2009 when EO-1 Hyperion was declared free. This has motivated some studies to
make first step,of ambitious project: tropical biodiversity mapping. As tropical country with rich biota, biodiversity mapping is very
important in Indonesia, a terrestrial spectrometry completed the data for deeper analysis. A biodiversity map will be derived from
EO-1 Hyperion data over Bogor Botanical Garden. The area covers about one kilometer square. The study shows that there is a need
for building better tropical spectral library.
1. INTRODUCTION
Biodiversity map is an important issue for todays global earth
condition. It is not just an interest for biologists and
environmentalists,but now it deals with the global issue of
climate change, where biodiversity richness in every part of the
earth is under threat for extinction. Biodiversity (richness)
occurs at many scales and includes genetic diversity, species
diversity (species richness), functional diversity, and ecosystem
diversity (Gamon, 2008). Tropical ecosystems are among the
world’s hotspots of species richness and endemism. Many of
the forested ecosystems are disappearing or being degraded at
rapid rates. In the past, remote assessment of biodiversity has
been largely indirect because satellite sensors have generally
lacked the spectral and spatial resolution needed to capture
patterns of species richness directly. Additionally, optical
remote sensing of biodiversity is generally limited to what can
be detected from above and is heavily weighted by the signal
returning from top surface layer - for example, the upper
canopy layers of a forest.
With the advent of new imaging spectrometers, we now have
several more direct pathways for linking remote sensing to some
measure of biodiversity. Some of these alternate measures of
diversity may be more accessible from remote sensing than
species richness. When combined with spatial detail (small
pixel sizes) and temporal resolution (multitemporal imagery),
hyperspectral sensors provide a rich array of tools for direct
assessment of vegetation functional and structural diversity
from remote sensing.
The use of remote sensing to detect biodiversity depends on the
biodiversity level being measured and this relates to either
spatial or spectral resolution. If there is high correlation
between remote sensing and biodiversity then there will be a
strong correlation between the optical diversity with the surface
canopy or other measures of biodiversity detectable.
Hyperspectral data are narrow band information of the
reflectance of objects on the earth surface. The data are usually
acquired from satellite, aircraft, or from a spectroreadiometer.
This kind of technology provides advantages in analising the
spectral object information in detail. One of the spectral data
that are still under research stage is the Hyperion, carried by the
EO-1 satellite (Nemani et a,l 2003). Hyperion data contain 242
spectral band that can be classified into AVNIR spectral range
(bands 1-70) to SWIR spectral range (bands 71-242). There are
220 unique spectral bands or channels in Hyperion images, with
a complete spectrum covering from 357 to 2567 nm. From the
242 bands contained in Hyperion data only 198 bands were
calibrated, and the uncalibrated channels were due to the
detectors’ low responsivity. These are the level 1 Radiometric
product. There is an overlap between the VNIR and SWIR so
that only 196 bands are unique. The calibrated channels are
band 8 to 57 for the VNIR and 77-224 for the SWIR
(University of Cincinnati, 2003).
In general, the capability of hyperspectral sensor to detect many
narrow spectral bands can be applied to detect chemical and
anatomic characteristics from many plant reproductivity. Some
studies have shown the advantages of using narrow bands
(hyperspectral data) using certain spectrum in comparison to
wide spectral band (multipectral data) for example, to obtain
quantitative and qualitative information of the most sensitive
part in vegetation or crop plants (Elvidge, 1990; Adams et al,
1995; Gao, 1996, Ceccato et al, 2001).
Because of different spectral width used by hyperspectral data,
this requires different approaches in its processing procedures.
For mapping purposes, the commonly applied project is the
Corresponding author.
