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
Figure 1. HJ-1 CCD and Radarsat-2 image of study area, (a) HJ-1 CCD (RGB normalization false color composition), (b)
Radarsat-2(PauliRGB composition)
Data source Acquisition data | Spatial data(m) Signal characteristic Other notation
HJ-1 CCD 2010-08-04 30x30 spectral reflectance four band number
RADARSAT-2 2010-08-01 12x8 backscattering coefficient | fine quad-polarization
Table 1. Overview of RADARSAT-2 image and HJ-1 CCD image
2.3 Situ measured data
Coincident with the image, sea surface measurement data,
including TIN and ocean surface condition (i.e., wind speed,
wind direction, wave height and wave direction) were collected
from Environmental Protection Department, the government of
the Hong Kong Special Administrative Region
(http://sc.epd.gov.hk).
3. METHOD
3.1 Selection of input variables
In order to improve the accuracy of estimating TIN, the input
variables which were relative sensitive to TIN were selected.
To ascertain the sensitivity of the spectral parameters and
polarization parameters to TIN in the study area, a series
quadratic fit model were conducted from TIN value with
spectral parameters and polarization parameters. In general, for
polarization parameters, our results showed that HV has a
06r
d os. Ra
E i T t
# 0.4 "st E ose 2, io
TIN(mg/L)
= n
Sin
EFA n £x
Qum
1 : ;
6.00 001 02 0.03 (£04 005
Band4
Figure 2.
Relationship between different optical parameters, polarization parameters and TIN
strong non-linear relationship with TIN, the correlation
coefficient (R?) being 0.54; VH correlate positively to TIN, R°
with 0.36 (Figure 2).
For spectral parameters, Normalized Difference Spectral Index
(NDSI) was proposed and computed by
NDSI - (R, - R)/ (R, - R) (1)
Where R, and R, are reflectance of 1” band and
reflectance of 4™ band for HJ-1 CCD image. Our results
indicated that single band, namely, reflectance of Band 4 was
positively correlated with TIN, R? of 0.35; NDSI has
correlation with TIN, R? with 0.32 (Figure 2).
In sum, VH, HV, Band 4 and NDSI were selected as the input
variables of model for estimating TIN in sea surface.
4 AL [0 4.2 64
NDSI