h 
lems 
face 
: 
cing, 
sets 
g 
sider 
"istic 
ining 
area 
uired 
tance 
tance 
force 
1ality 
h are 
ance. 
depth 
tance. 
»vides 
; been 
isition 
dcover 
2004). 
these 
sources 
ber of 
hat are 
ig. from 
feature 
hod for 
yield a 
ntext of 
ate, the 
quently 
physical 
sive. A 
features 
er areas 
ty and 
pogenic 
International Archives of the Photogrammetry, Remote Sensing 
origins, alter spectral properties of benthic cover. Hence there is 
a need to consider these factors in understanding images 
obtained from different sources taken at various periods. 
2. SITE, DATA AND PROBLEM APPROACH 
The goal of synergy is to reconcile all imagery into a common 
resolution without loss of detail necessary to accomplish 
classification and feature tasks. The first step is to convert 
nominal radiance values measured by each satellite sensor into 
a single standardized reflectance image dataset with the basic 
criteria that corrections must be rooted on theoretical grounds. 
Reflectance detected by each sensor is rectified according to 
acquisition date and time, imaging geometry: these two 
acquisition conditions regulate incoming radiant energy and 
modulate reflectance. Since water depth governs light 
penetration, tide level conditions are accounted for carefully as 
it changes even among images acquired by a single sensor. 
Coastal habitats are further complicated by the dynamic nature 
of the water column. Optically-active constituents present in the 
water modify tone and pigmentation are therefore modelled. To 
compare reflectance values from satellite sensors with that of 
in-situ instruments such as field spectroradiometers, the spectral 
response functions of the sensors per band range are taken into 
account. 
2.1 Target sites and field spectral measurements 
There are two areas selected for this study: Shiraho Reef 
(24°21’N, 124°15’E) and Fukido River mouth area (24°29°N, 
124?13'E). Both are located in Ishigaki Island, Southern 
Ryukus. Shiraho Reef typifies a subtropical fringing reef 
(Figure 1) and abundant with almost 200 coral species. On the 
other hand, the moat area reef area outside Fukido River is 
notable for mixed seagrass beds in abundant cover. Five cross- 
shore (5) transect lines in Fukido River mouth and another five 
(5) Shiraho Reef area were established. The average length of 
the transects in Fukido is about 200 meters while the transects 
in Shiraho is approximately 400 meters. Depths along these 
transects reaches 25 meters but are about 1-3 meters on the 
average. The cross-shore transect configuration allows for 
sampling of all possible shallow benthic cover on the two types 
of shallow reef areas. 
Onboard a small motor boat, a portable dual-channel 
Spectroradiometer (Ocean Optics S2000®: spectral resolution 
2058 pixels equivalent to 332 nm to 1016nm range) was used to 
gather horizontal spectral profiles along the transects. The 
sensor of one channel is equipped with a glass diffuser and 
focused vertically upward. We designate this as the sun sensor. 
On the other hand, the sensor of the other channel (object 
sensor) directed vertically downwards, is submerged and 
attached to a buoy. The buoy is left to float about 4 meters 
away from a boat so the latter casts no shadow over the bottom 
cover being measured by the former. The boat is then driven 
along the transects while a portable computer continuously logs 
the spectral data channel and location from a GPS (Global 
Positioning Systems) unit at 5-km/hr average speed. At 300ms 
This speed enabled acquisition of spectra at 3 scans per meter. 
Ground truth data were collected for the same transect lines 
where the spectral measurements were obtained. Field data 
obtained include position, depth (both by echosounder and line 
Measurements), benthic cover, coral, seagrass and algae 
distribution in Jul 2002 and Aug 2003. Above and below-water 
PAR sensors were used to measure in-water solar irradiance. 
997 
and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
JAPAN 
Muyang 
East China Sea OKINAWA 
Fuzhou : 
Naha’ 
,.Chilung RYUKYU ISLANDS 
Hsin-chu 4 Taipei 
4 
Taichung y Pi North Ryukyu 
: E. Trench 
Cha Kuang: Huadien Yaeyama 24° 
Islands 
TAIWAN 
) .” Taitung PACIFIC OCEAN 
Rushing Philippine Sea 
120 124° 126° 128° 
£7 
: © 
Hirakubo-saki 
East China Sea 
24° 30' 
  
     
24° 30 
o 
e 
à EE 
YAEYAMA X. Ishigaki 
NS SHOTO Island. 
Promos S =f PACIFIC 
S -Jima + 4 & OCEAN 
Sekisei Taketomi 
Lagoon 
E. Philippine 
v Kuroshima PI 
Sea 
Figure 1. Location of field sites for this study. Above: vicinity 
of Ishigaki Island. Below: upper block show Fukido area, lower 
rectangular block covers Shiraho Reef area. 
2.2 Satellite images 
  
IKONOS MS 
  
IKONOS Pan 
  
ASTER 
SPOT 5 XS 
Satellite sensors 
SPOT 5 PAN 
Landsat ETM+ 
  
   
Landsat ETM+ PAN 
  
Wavelength (nm) 400 500 600 700 800 900 
Figure 2. Summary of spectral sensitivity ranges of sensors 
used for this study. 
One scene from each of the multispectral sensors of SPOT XS 
(processing level: Level 2A; spatial resolution: 10m), Ikonos 
(Geocorrected; 4m), one Landsat ETM+ (2b; 30m) and ASTER 
(Ib: 15m) images were used for this study. Figure 2 illustrates 
band span ranges of these sensors while Table | describes the 
date, time of acquisition and observation geometry of images. 
All images were acquired in first quarter of 2002 to reduce 
seasonal variations across images. Atmospheric correction was 
applied to all the images by the applying the SPECTRL2 model 
(Bird and Riordan, 1984) using the solar data from the Ishigaki 
meteorological station with the same atmospheric column 
conditions (ozone, aerosol optical depth and water vapour), 
reducing the values to surface reflectance at sea surface. Subset 
of the images covering the two target areas (Shiraho: 35 sq km.; 
Fukido: 4.5 sq. km) were then segregated for further processing.