/, 2012 
14(10): 
oth 
) Tree 
Caartinen 
tion from 
10): 
and 
e 
ication. 
ng. 
T and 
r a boreal 
, lidar and 
'orest 
ar 
ing of 
  
  
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 
FUSION OF OPTICAL DATA AND SAR DATA FOR THE ESTIMATION OF NITROGEN 
CONCENTRATION IN PEARL RIVER ESTUARY HONG KONG SEAS, CHINA 
Xiangnan Liu*, Meiling Liu, Ling Wu 
School of Information Engineering, China University of Geosciences, 100083 Beijing, China liuxn@cugb.edu.cn, 
liumeiling427@126.com, wl_19830807@sohu.com 
KEY WORDS: Total inorganic nitrogen, RADARSAT-2, HJ-1 CCD image, backscattering coefficient, optical parameters, 
polarization parameters 
ABSTRACT: 
The knowledge of nitrogen concentration in the ocean is fundamental for the study of oceanic biogeochemical processes. The 
objective of this research is to estimate total inorganic nitrogen (TIN) by integrating optical parameters from HJ-1 CCD image and 
polarization parameters from RADARSAT-2 quad-polarization image. The situ data and HJ-1 CCD, RADARSAT-2 image were 
acquired from Pearl River Estuary Hong Kong Seas, China in August, 2010. The four sensitive parameters, reflectance of Band 4, 
NDSI (Normalized Difference Spectral Index), the backscattering coefficient of HV and VH were derived as input variables to assess 
the TIN. A multiple regression model was established between four input variables and TIN. The result showed that the fusion of 
optical data and SAR data was proved to be successful in estimating TIN in sea surface, with the correlation coefficient (R?) between 
measured TIN and predicated TIN of 0.774, and the root mean square error (RMSE) of 0.063. The optical data in combination with 
SAR data is promising for detecting biochemical component in sea surface. 
1. INTRODUCTION 
The knowledge of temporal and spatial variations of nitrate 
concentrations at global or regional scales in the ocean is 
needed to quantify the role of nitrates in oceanic 
biogeochemical processes, and in particular those linked to new 
primary production (Kamykowski et al., 2003, 2005). Many 
attempts have been made to estimate nitrate concentrations 
using satellite data. Some researchers applied satellite data to 
estimate nitrate concentrations in sea according to 
temperature—nitrate relationships based on matching vertical 
profiles of sea surface temperatures (SST) and nitrate 
concentrations (Traganza et al., 1983; Dugdale et al., 1989). In 
order to increase the accuracy of estimation nitrate 
concentrations, other researchers derived inverse relationships 
between nitrate concentrations and chlorophyll-a concentration, 
SST by introducing chlorophyll-a concentration as an additional 
input to empirical algorithms (Goes et al.,1999,2000; 
Silio-Calzada et al., 2008). However, the empirical algorithms 
derived from the chlorophyll-a, SST were limited to restricted 
periods and areas, due to varying hydrodynamics and 
biogeochemical characteristics condition. 
In this paper, we proposed a new approach for the estimation of 
nitrate concentrations in sea surface, by deriving the sensitive 
remotely sensed parameters from optical image and SAR image. 
2. STUDY AREA AND MATERIAL 
2.1 Study Area 
The study area (22?18'08"N, 114?03'30"E) is located in the east 
of the Pearl River Estuary (PRE), China, which is 
well-developed economical district and has many industrial 
  
* Corresponding author: e-mail: liuxn@cugb.edu.cn 
     
operations, which polluted the coastal areas. The site has a 
warm and humid subtropical climate, the Pearl River flows 
through large catchment areas into the PRE and finally reaches 
the South China Sea, which is the largest marginal sea on the 
western boundary of the Pacific Ocean. 
2.2 Image data 
A scene of HJ-1 CCD image was acquired on 4 August 2010 
(Figurel(a))HJ-1 satellite equipped with a CCD camera and 
hyperspectral imager (HSI) or infrared camera (IRS), which was 
successfully launched on at 11:25 on September 6, 2008,China 
to monitor environment and disaster. The HJ-1 CCD image has 
a four spectral band (Bandl: 0.43-0.52um, Band2:0.52-0.60um, 
Band3:0.63-0.69um, Band4: 0.76-0.90um) with a 30mx30m 
spatial resolution. The two identical CCD cameras in the 
HJ-1-A satellite and HJ-1-B satellite can image the ground 
swath width of 700 km. Revisit cycle is two days. The 
preprocessing of HJ-1CCD imagery includes atmospheric 
correction and geometric correction. 
In this study, a scene of fine quad-polarization RADARSAT-2 
image covering the study area was acquired over the study area 
on 22:28, 1 August 2010 from China Remote Sensing Satellite 
Ground Station (Figurel (b). Band beam is Q19, spatial 
resolution is 12mx8 m, incidence angle is about 22° .The image 
has been preprocessed to product level of SGX and output to 
GeoTIFF data format. The Next European Space Agency (ESA) 
SAR Toolbox (NEST) was used for further data processing, 
including radiance calibration, geocorrection, coregistration, 
multlooking and speckle reduction. For speckle reduction, the 
multitemporal speckle filter was adopted, and the filter window 
size was set 3 x 3.Overview of RADARSAT-2 image and HJ-1 
CCD image are displayed in Tablel.