Shallow Water Bathymetry of Singapore’s Highly Turbid Coastal Waters: A Comparative Approach
James F. Bramante, Durairaju Kumaran Raju, Sin Tsai Min
Tropical Marine Science Institute, National University of Singapore
Purpose
• Determine effectiveness of multispectral algorithms in Singapore
• Determine how extra 4 bands may help
• Develop high resolution shallow-water bathymetric map
• Coral/Benthic Surveys• Interface into more complicated IOP models• Determine possible new benthic habitats
Study Area
Study Area (cont.)
Marine EnvironmentStudy Area (cont.)
Wild Singapore
Seagrass-watch
Pulau Hantu
Obstacles
• High turbidity
• Sediment plumes
• Few bathymetric data points in shallow waters
Obstacles (cont.)
• High shipping traffic
• Abundant clouds
• Mixed aerosols from city and ocean
Atmospheric Correction• Stock image (no concurrent field measurements)
• Access to atmospheric information limited
• Clear boundaries for cloud, shadowed, and deep ocean pixels
Atmospheric Correction (cont.)
• Cloud-shadow empirical algorithm• Reinersman et al. (1998) and Lee et al. (2005)
Fig. 1 taken from Reinersman et al. (1998)
Atmospheric Correction (cont.)
Dws
pathtw t
LLL
• General Equation:
• Cloud-Shadow Eq:
• Assumptions:
• Lee et al. simplification:
shpathnepath LL
nersdnewsnepathne REtLL
shrsdiffshwsshpathsh REtLL
shwsnews tt shrsners RR
Atmospheric Correction (cont.)
d
diff
shnenepath
E
ELL
LL
1
)(• Path radiance:
• Reflectance:
• Water-air boundary correction:
rsCloudpathcloud
pathtrs R
LL
LLR
)(
rs
rsrs r
rR
562.11
518.0
Atmospheric Correction (cont.)
0.05
0.1
0.15
0.2
0.05
0.1
0.15
0.2
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
Band 2 (Blue) Average Radiance
Band 2 Atmospherically Corrected Reflectance
Bathymetry Algorithms
• LUT Classification
• Linear Ratio Algorithm (Stumpf et al. 2003)
• Linear Band Algorithm (Lyzenga et al. 2006)
• Compared results using conventional 4 bands and Worldview-2’s 8 bands
Bathymetry Algorithms (cont.)
• LUT Classification• LUT Library n = 53 for 0 < depth ≤ 2 m• Least squares comparison• Attempted ratio classification
8-band 4-band
Bathymetry Algorithms (cont.)
• Linear Ratio• Based off of Beer’s law:
• Stumpf et al. 2003 : )ln(
)ln(ˆ1
rsHI
rsLOo nr
nrhhh
hWBWrs errrr )(
Bathymetry Algorithms (cont.)
• Linear Band• Lyzenga et al. 2006 :
• Non-real results when LWCj > Lj
)ln( WCjjj LLX
N
j jjo Xhhh1
ˆ
Results
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2Lyzenga et al. Algorithm Band Classification
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Results (cont.)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Lyzenga et al. Algorithm Band Classification
Results – Platform Comparison
Results – Faulty Relationships
Conclusions
• With more validation, Lyzenga et al. model and band classifications may prove useful in turbid waters
• Assumed relationship between band absorption and depth must be re-examined in extremely turbid waters
Further Investigations• Evaluated cloud-shadow atmospheric correction model against RT Model; former was validated and did not affect results much
• Attempting to use water-column index to adjust for water mass variation in Lyzenga algorithm
• Using spectroradiometer to modify semi-analytical models for Singapore