Contribution form remote sensing
in updating bathymetric chart
K. Abdullah , M. Z. MatJafri and Z. B. Din
Universiti Sains Malaysia
11800 Penang, Malaysia
Tel: 604-6577888 Fax: 604-6579150
E-mail: khirudd@usm.my, mjafri@usm.my
Keywords : algorithm, bathymetry, remote sensing.
Abstract
The study area selected is the Strait of Penang, located in the north western part of peninsular Malaysia. Surveys were conducted to measure new sounding points using a boat equipped with an echo sounder. The station locations were determined using a GPS system. The data were analysed and tidal corrections were applied to reduce the measured water heights to the chart datum. The tidal readings obtained from the nearest station were used for reference in the data analysis. A new suggested bathymetric chart was drawn based on the new sounding points. We attempted to employ satellite imagery as an aid to bathymetric charting. Cloud free scenes of Landsat TM and SPOT data acquired between January 1997 and February 1997 were available for use in the present study. Each subscene was rectified to the corresponding bathymetric chart. Image locations were related to map GCP coordinates through the second degree polynomial transformation equations. The pixel values at the sample locations were extracted and used as independent variables. Multiband water depth algorithm was used in the calibration analysis. Regression technique was used for calibration of the satellite signals for water depth measurement using the measured sounding points as dependent variables. The correlation coefficient and root-mean-square deviations were examined for each data set. The accuracy of each calibrated algorithm was further verified using other known sounding points. The respective calibrated algorithm was then applied to the coresponding image to generate a water depth map. The colour coded bathymetric map was used for visual interpretation and assessment of the present water depth pattern.
Introduction
Although the existing bathymetric charts were reprinted recently, they display water depth values which were measured decades ago. Oceanographers have often questioned the reliability of these charts. Costal erosion and siltation as well as activities like dredging and land reclaimation can cause deposition and erosion of the sea floor and hence changing the bottom topography. Therefore a resurvey of the present water depth pattern was conducted in an attempt to update the existing bathymetric chart.
In this present study the use of both the algorithms (using single-band, two-band ratio and three?band) and the proposed three-band methods were investigated in coastal areas of Malaysia. The aim of the study is to assess the accuracy provided by different types of satellite data in bathymetric mapping. TM and SPOT data were used in the present comparative analysis. The limitations faced by the remote sensing technique are discussed.
Physical Principles of Water Depth Determination
The development of water depth determination techniques was based on the simple water reflection model which accounts for the major part of the signal received by a sensor over a clear shallow water (Lyzenga (1978)). According to this model, the radiance, L, in a single wavelength band can be written as
L = Ls + LoRbexp(-KfZ) (1)
where L
o is a constant which includes solar irradiance, the transmittance of the atmosphere and the water surface and the reduction of the radiance due to refraction at the water surface; L
s is the radiance observed over deep water (due to external reflection from the water surface, volume reflected radiance from the water column and scattering in the atmosphere); R
b is the bottom reflectance; K is the effective attenuation coefficient of water; f is a geometrical factor to account for path length through water and Z is the water depth. Other models developed by Benny and Dawson (1983), Stove (1985), Philpot (1989), and Lee et al. (1999) were also studied.
Regression Algorithms
As far as the regression is concerned the expression should be in the most simplified form. The single-band depth algorithms derived from the above model can be written in a common simplified form for channel i
Z = aoi + a1iXi (2)
where X
i = ln(L
i - L
si), a
oi = (1/K
if)ln(L
oiR
bi) and a
1i =-1/K
if
The above equation can be regressed linearly with in-situ water depths. The limitation with a single band method is that changes in the bottom reflectance within the area of interest will cause errors in the depth calculation. Predes and Spero (1983) derived the n-band/n-bottom type case as
Z = a0 +a1X1 + a2X2 +…+ anXn (3)
We modified the above algorithm by writing equation (2) for 3 bands as simultaneous equations and finding the single expression for z for the 3-band case. The form of the series obtained:
Z = ao +a1X1 +a2X2 +a3X3 +a4X1X2 +a5X1X3 +a6X2X3 +a7X1X2X3 +… (4)
For the present bathymetric application we used the series up to the term displayed above and the coefficients were determined by regression.
Data for the Area Under Investigation
The study area is the shallow water region in the Straits of Penang, Malaysia (between latitudes 5.30° to 5.45° N and longitudes 100.30° to 100.40° E). Images scenes of SPOT and TM captured on 1 January 1997 and 28 February 1997 respectively, were selected for analysis. The bathymetric chart of Penang (chart 1366, 1:60000 scale, published at Taunton, England, revised in 1999) was used for reference. New depth sounding data measured using echo sounder were available. Tidal data corresponding to the scenes were acquired. Data on water quality parameters and other oceanic parameters were not available for reference.
