SYNERGISTIC METHODS IN REMOTE SENSING DATA ANALYSIS 
FOR TROPICAL COASTAL ECOSYSTEMS MONITORING 
2 (^ ssp 8, Dy 4 ë 
E. C. Paringit ^ "*, K. Nadaoka" 
? Dept. of Mechanical and Environment 
2-12-1 Ookayama, Meguro-ku, 152-8552, Tokyo, Jay 
? Dept. of Geodetic Engineering, University of the Philig 
KEY WORDS: Ecosystem, Inte 
ABSTRACT: 
Tropical coastal environments around the world have undergone rapid ¢ 
ts to map their extent and distribution from sp 
degradation inevitable. Despite numerous attemp 
signals reflected and subsequently measured by ren 
have been given inadequate attention. The dynamic c 
water quality and environmental stresses, both natur 
these factors in understanding images obt 
methods in multi-source image processing 
image data covering these types of environment from 
covers the Fukido river mouth area and Shiraho reef of Ishigaki Isl 
from satellite-borne Ikonos, SPOT, ASTER and Landsat respectively in 2002. 
ow water optics and radiative transfer 
as distribution, abundance, morphology an 
ational results, field surveys were conducted to gathe 
face conditions, benthic habitat cover, abundance and d 
iles across the reef area benthic cover were 
distribution functions) modelling, shall 
values with biophysical properties such 
parameterizations and to validate comput 
(mainly chlorophyll-a and turbidity), sea sur 
synchronous with image acquisition. Spectral prot 
The developed reflectance model was appli 
and benthic cover estimates. Results showed relative proximity of 
source image for the same area are 
The accuracy of the cover and depth estimates satellite sensor 
a physical basis for relating differe 
1. INTRODUCTION 
1.1 Background on Remote Sensing of Coastal Habitats 
ent global climate anomalies and increased 
Due to recurr 
marine environments around 
habitation in coastal zone, tropical 
the world have undergone abrupt and undesirable changes. 
Reckless utilization of coastal resources resulted in 
deterioration of their nurtured habitats (coral reefs, seagrass 
meadows and mangrove stands). In order to expediently devise 
proper conservation measures and formulate sustainable 
management alternatives for these coastal ecosystems, there is a 
ans to obtain reliable information on the 
need first to develop me 
state of their health and well-being, and thereafter provide tools 
to continuously monitor them. 
and distribution of coastal marine 
numerous and are already near 
has been confined merely for 
on a piecemeal and 
Attempts to map the extent 
habitats from space data are 
pervasion. Activities, however 
mapping shallow benthic coverage 
intermittent mode. Regardless of restrictions cost and weather 
conditions, the application of conventional remote sensing 
analysis approaches to any single satellite data (c.g. IKONOS, 
Landsat TM, SPOT) in current operation barely go beyond 
classification accuracy above 70% (Mumby and Edwards., 
2002). 
EL 
* Corresponding author. 
note imaging sensors to the bioph 
haracteristic of the coastal shallow wate 
al and anthropogenic complicate this t 
ained from different sources taken at v 
for assessing benthic coastal habitats such 
space through the aid of theoreti 
ed to the image datasets by uti 
al Informatics, Tokyo Institute of Technology, 
san — ecp@wv.mei.titech.ac.Jp, nadaoka@mei.titech.ac.jp 
spines Diliman, Quezon City, 1101- ecp@engg.upd.edu.ph 
gration, Marine, Multisensor, Multitemporal, Multispectral, 
hanges which made consequent of their ecosystems 
ace, the ability to relate the surface 
ysical characteristics of coastal habitat targets 
r areas including tidal and wave forcing, 
ask. Hence there is a need to consider 
arious periods. This research focuses on synergistic 
as corals, seagrass, and algae by examining 
ical remote sensing approaches. Our study area 
and located in southern Ryukus, Japan. Images were acquired 
Principles of BRDF (bidirectional reflectance 
have been utilized to explain shallow water reflectance 
d depth as controlling parameters. To reinforce 
r in-situ data including water quality 
istribution, some of which are 
also used for calibrating reflectance. 
lizing model inversion techniques, hence obtaining depth 
image-derived reflectance to processed in-situ spectral reflectance. 
also presented. The model provides 
nt image datasets from different sources. 
In terrestrial and global fields, immense interest has been 
devoted in taking advantage of the repetitive acquisition 
capability of remote sensors for discriminating landcover 
features and for detecting associated changes (Coppin, 2004). 
Over tropical coastal habitats, the ability to combine these 
sources and make inferences from a multitude of image sources 
are met with immense challenge due to a number of 
considerations inherent to sensor systems and those that are 
attributed to the nature of coastal environment. 
1.2 The need for synergistic approach 
It is hypothesized that combination of images coming from 
various sources may lead to improved performance of feature 
extraction and classification. This paper outlines a method for 
combining imagery from different sensors that would yield a 
compatible product useful for processing them in the context of 
extracting resource information in coastal zones. To date, the 
ability to relate the surface signals reflected and subsequently 
measured by remote imaging Sensors to biophysical 
argets remain elusive. À 
characteristics of coastal habitat t 
compounding difficulty in spectrally res 
is that the dynamic nature of the coasta 
including tidal and wave forcing, 
environmental stresses, both from natura 
olving habitat features 
| shallow water areas 
water quality and 
| and anthropogenic 
996 
Int 
ori 
obt 
Th 
res 
cla: 
non 
à s 
crit 
Rel 
acq 
acq 
mo 
pen 
it € 
Coa 
of tl 
wal 
com 
in-s. 
resp 
acce 
2.1 
Thei 
(24? 
124° 
Ryu 
(Fig 
othe 
nota 
shor 
(5).« 
the t 
in S 
trans 
aver: 
sam[ 
of sh 
Onbe 
speci 
2058 
gathe 
sensc 
focus 
On t 
Sensc 
attacl 
away 
cove 
along 
the s 
Positi 
This : 
Groui 
wher 
obtaiı 
Meası 
distril 
PAR