International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
SENSITIVITY ANALYSIS IN THE RETRIEVAL OF TURBID COASTAL WATER 
BATHYMETRY USING WORLDVIEW-2 SATELLITE DATA 
S. C. Liew*, C. W. Chang, L. K. Kwoh 
Centre for Remote Imaging, Sensing and Processing (CRISP), National University of Singapore, 
Blk. S17 Level 2, Lower Kent Ridge Road, Singapore 119076 
(crslsc, crscew, crsklk,)@nus.edu.sg 
Commission VII/1 
KEY WORDS: Bathymetry, turbid waters, ocean color 
ABSTRACT: 
The recently launched Worldview-2 satellite provides high resolution (2-m multispectral) data in eight spectral bands in the visible 
to near-infrared region. The additional spectral bands provide an opportunity to test several algorithms for retrieving the water depth, 
bottom albedo and intrinsic optical properties of coastal sea water. In a previous work (Liew et al., 2011) we reported our attempts in 
retrieving water depth and bottom albedo using WorldView-2 data for the purpose of coastal habitat mapping. In this paper, we 
investigate the sensitivity and limitations in using WorldView-2 spectral bands for bathymetry retrieval in turbid coastal waters. For 
typical coastal waters with a dark seabed, the most sensitive band is the Green Band which is sensitive to water depth up to about 5.3 
m. For coastal waters with a bright sandy seabed, the Red and Yellow Bands are the most sensitive, but the maximum sensitive depth 
is reduced to about 2.4 m. 
1. INTRODUCTION 
High spatial resolution satellite sensors are usually designed for 
land applications. These sensors typically have a small number 
of broad spectral bands. Ideally, hyperspectral data are 
preferable for retrieving water optical properties and water 
depth (Lee et al, 2002) as the complete water reflectance 
spectra are available for fitting with specific models of water 
reflectance. Despite the limited number of spectral bands, 
attempts have been made in deriving the intrinsic optical 
properties, water depth and sea bottom albedo using data from 
high resolution satellites such as the SPOT-5 satellite (Liew and 
He, 2008) which has 4 spectral bands in the green, red, near 
infrared and short-wave infrared regions. The recently launched 
WorldVview-2 satellite sensor has 8 spectral bands and holds 
the potential to be used for deriving optical properties and 
bathymetric maps from littoral zone waters. In our previous 
work, bathymetric and coastal habitual maps were derived from 
WorldView-2 images (Liew et al., 2011). 
For deriving the optical properties of water, most ocean colour 
satellite sensors have spectral bands with a narrow bandwidth of 
about 10 to 20 nm. The spectral bands are selected at different 
positions to pick up absorption signatures of phytoplankton and 
to overcome the non-linear relationships between ocean colour 
and optical properties of various constituents. It will be of 
interests to evaluate possible limitations due to the broad 
bandwidth in deriving optical properties and bathymetric maps 
using data from the WorldView-2 satellite sensor. 
In this paper, we used well-known semi-analytical equations 
(Lee et al., 2002) to simulate the remote sensing reflectance of 
coastal waters at high spectral resolution. The water leaving 
radiance spectrum was computed and aggregated into the 
effective in-band radiance for each spectral band of the 
  
* Corresponding author. 
13 
WorldView-2 sensor using the sensor’s relative spectral 
response functions. The in-band radiance of each spectral band 
was converted to the effective reflectance. We then performed 
sensitivity analysis of the effective reflectance of each spectral 
band with respect to the change in water depth. 
2. WORLDVIEW-2 SENSOR 
The WorldView-2 satellite sensor has 8 multispectral bands at 
1.84 meter resolution. The average spectral bandwidth is about 
50 nm for the first six bands and about 100 nm for the last two 
near-infrared bands. The relative spectral response curves of the 
8 spectral bands are shown in Figure 1. The wavelengths and 
bandwidths are listed in Table 1. 
3. SHALLOW WATER REMOTE SENSING 
REFLECTANCE MODEL 
The remote sensing reflectance just below the water surface is a 
sum of two components (Lee et al., 1998; Lee et al., 1999), 
(A) 
ns) n, QJ -exp(-MKH)]- PE exp(-MKH) (1) 
TT 
The first component is due to scattering from the bulk water 
characterized by the deep water reflectance 5,(A) related to 
the water absorption and backscattering coefficients, a(X) and 
bp(A) and two constant parameters gp and g; via the 
equations (Gordon et al. 1988), 
KM) =Lg0 + £140)]4(.) , (2) 
bp(A) 
A eee 
20) a(4.) *- by (X) ©)