VNU Journal of Science, Earth Sciences 25 (2009) 217-222

' .. \' ~

Application of remote sensing for shoreline change

detection in Cuu Long estuary

Tran Thi Van 1•*, Trinh Thi Binh2

1
Institutefor Environment and Resources, Vietnam National University Ho Chi Minh City
'University ofNatural Science, Vietnam National University Ho Chi Minh City
Received 2 November 2009

Abstract. Coastal zone of Cuu Long estuary is a place through which Mekong River flows into
the Eastern Sea with 8 estuaries. is the zone is formed under the influences in the interaction
between river and sea. Geologically, coastal zone of Mekong delta is characterised by the
predominat of clay and silt formations. These two factors makes the shoreline sensitive to
exogenous processes such assedimentation and erosion .. This paper presents an application of
satellite remote sensing technology to detect and analyze the spatial changes as well as quantify
the shoreline change in Cuu Long estuary. Landsat and Aster satellite images were used with band
ratio method for shoreline change detection.The results present shoreline changes maps in three
periods: 1989, 2001 and 2004. The outcomes of the case study can be used as an orientation for the
sustainable integrated management plan of coastal zones.
Keywords: accretion, band ratio, erosion, change detection, shoreline.

1. Introduction detecting shoreline changes can be roughly

divided into four categories, all of which have
Shoreline, the boµndary )Jetween land and both advantages and disadvantages: (I)
sea keeps changing its shape and position conventional ground surveying can achieve
continuously due to dynamic conditions. The high accuracy of measurement, but is labor
change in shoreline is mainiy associated with intensive and time consuming; (2) modem
waves, tides, winds, periodic storms, sea-level altimetry technology uses radar altimeters or
change, 'the geomorphologic~l processes such laser altimeters. It has a great potential, but the
as erosion and accretion and human activities detectors are currently less available; (3)
[1]. Erosion and accretion affect human life, airborne imagery measurement provides
agricukture and aquculture practice and sufficient pictorial information, but the
waterway transport activities: Detection and frequency of data acquisition is low, and the
measurement of shoreline changes are an photogrammetric procedure including data
important task in environmental monitoring and acquisition and image mapping is costly as well
coastal zone management. Approaches to as time consuming [2]; (4) Multispectral remote
sensing satellites provide digital imageries in
'Corresponding author: Tel.:84-919188485 infrared spectral bands where the land-water
E-mail: [email protected] interface is well defined. Furthermore this
217
218 T.T. Van, T.T. Binh I VNU Journal of Science, Earth Sciences 25 (2009) 217-222

method has advantages: not time consuming, to April. The river system with high density is a
inexpensive executed cost and large ground place through which Mekong River flows into
coverage monitoring [2,3]. the Eastern Sea with 8 estuaries.
Satellite optical images are simple · to
interpret and easily obtainable. Absorption of
infrared wavelength region by water and its
strong reflectance by vegetation and ~oil make
such images an ideal combination for mapping
the spatial distribution of land and water. These
characteristics of water, vegetation and soil
make the use of the images that contain visible
and infrared bands widely used for coastline
mapping [4]. This study applied satellite image
interpretation and GIS to detect and analyze the
spatial changes as well as quantify the result of
shoreline change in Cuu Long estuary. The
results present shoreline change· maps in three
periods: 1989, 2001 and 2004.
Fig. I. The study area.

2. Study area and data sets 2. 2. Data set

2.1.Study area Satellite images were acquired in three
dates: Landsat TM in 1989, Landsat ETM+ in
Coastal zone of Cuu Long estuaries extends
from Tranh De river mouth to Tieu river mouth. 2001 and Aster in 2004 (Table 1). Landsat
Study area is located between latitude 105° 58' images have spatial resolution of spectral bands
58"N to 106°'51' 8" N and longitude 9° 23' 55" in 30m and ETM+ band 8 is panchromatic of
E to 10° 25' 37E (Fig. I). The mainland is l 5m resolution. Aster image has 2 groups of
limited by boundary of coastal districts, spectral bands with different pixel resolutions:
including 8 districts: Long Phu (Soc Trang VNIR bands in !Sm and SWIR bands in 30m.
Province), Cau Ngang, Duyen Hai, Tra Cu (Tra These image were taken in dry season.
Vinh province) Binh Dai, Thanh Phu, Ba Tri Therefore shoreline is easily identified due to
(Ben Tre province) and part of Go Cong Dong the fact that the water level is not very high.
(Tien Giang province). This area is relatively Topographical map at scale l: 100.000, UTM
flat with a lot of sand dunes turning backto the
coordinate and WGS-84 projection was use for
sea. The climate has tropic and monsoon
geo-rectifying and determining erosion/
characters with two seasons: the rainy· from
May to November and the dry from December accretion locations.

Table I. Satellite data used in the study

Acquisition Time Satellite Sensor Resolution

16 /01/ 1989 LandsatS TM 30m
11/12/2001 Landsat? ETM+ !Sm, 30m
12/12/2004 Aster Aster !Sm, 30m
 T.T. Van, T.T. Binh / VNU Journal of Science, Earth Sciences 25 (2009) 217-222 219

3. Methodology extract soil and water from Landsat and Aster

•.(mages.
Spectral band ratio is one of the most Image preprocessing: Aster image has the
common mathematical operations applied to smallest spatial resolution of 15m of VNIR
multi-spectral image data. Ratio images are bands, so band 4 (pixel size in 30m) in SWIR
calculated as the division of DN values in one spectrum will be downscaled to this one. This
spectral band by the corresponding pixel value process is also appliedto Landsat bands with
in another band. Band ratioing provides unique 30m 1 resolution. In change detection, geo-
information not available in any single band rectification of images with different spatial
that is useful for distinguishing earth surface resolution is constrained for guaranteeing the
features. Band ratio operation can reduce the smallest e1rnr in overlaying results. Aster 2004
environmentally induced variations in the DN was used to geo-rectify with the topographical
values of a single band, such as brightness map. Other Landsat images was rectified by
variations caused by topographic slope and Aster one. The RMSEs were less than 0.5 pixel.
aspect, shadows or seasonal changes in sunlight
Image analyzing: In the first step, histogram
illumination angle and intensity. Therefore,
threshold method is used on NIR band for
band ratioing can emphasize and highlight
separating land from water. The threshold
subtle variations in the actual spectral responses
values are selected so that all water pixels are
of various surface covers.
separated from the land pixels . In this case, few
Experiments have shown that green band in land pixels have been mistakenly assigned to
0.52 - 0.6µm wavelengths (Landsat band 2 and water pixels but not vice versa. Water pixels are
Aster band !) is sensitive to water turbidity then assigned to "I" and land pixels to "O''. A
differences plus sediment and pollution plumes, binary image has been achieved then. This
because it covers the green reflectance peak image is named "image I". In the second step,
from leaf surfaces. It can be useful for ratioing method is used with ratios green/NIR
discriminating broad classes of vegetation. and green/MIR. Green/NIR ratio is useful for
Water is strong absorber of near infrared separating land from vegetation, green/MIR for
radiation (NIR), so Landsat band 4 (0.76- separating non-vegetation land. Water pixels
0.90µm) and Aster band 3 (0.76-0.86µm) are will be greater than I. Two these ratio images
useful for locating and delineating water bodies, are multiply for rejecting mistake and forming
distinguishing between dry and moist soil and the new image named "image2". Image! is
providing information about coastal wetland, multiplied to image2 to generate third image
swamp and flooded areas. Landsat band 5 named "image3". Some isolated pixels are
{l.60-1.70µm) and Aster band 4 (1.55-1.75µm) remove with sieveing and filtering technique for
exhibits a strong contrast between land and the final shoreline image. The resulted shoreline
water features due to the high degree of extraction is transformed into vector format and
absorption of mid-infrared energy by water and exported into Maplnfo format for analyzing
strong reflectance of mid-infrared (MIR) by change of erosion/accretion areas. Change in
vegetation and natural features in this range. pairs are examined such as 1989-2001 and
The wavelength information is necessary in 2001-2004. Fig. 2 illustrates the steps of the
ratioing settings. This study used band ratioing method.
method in thes mentioned wavelength region to
 220 T. T. Van, T.T. Binh I VNU Journal of Science, Earth Sciences 25 (2009) 217-222

and < threshold

.
Image3
·····~'11\111.--!l'\!!I
= image l x,,,image2
,-, " '''"''"'"' '"

Fig. 2. Flowchart of extracting shoreline from satellite images.

4. Results and discussion accretion/erosion sections in 1989-2001 are

Accretion and erosion are two inverse continued in the next stage in 2001-2004 with
processes. Accretion increases the area of the higher and more rapid measure. Section in
agricultural, forest and fishing land, but yields confluence between middle Hau river and Quan
the sediment in the sea port, baring waterway Chanh Bo channel had eroded in 1989-2001 ,
transportation activities. While erosion causes but had no change in 2001-2004. This region
the loss of land, destroys constructed buildings, was examined with full vegetation cover in 3
houses, threatening human life. The analyzing satellite images. Lots of sections were
results has shown the erosion/accretion areas sedimented in 1989-2001, but they were
through time in fig. 3. seriously eroded after 200 I. It lengthened and
joined with previous erosion sections. It is
Shore is often eroded robustly in well-aired
obviously observed in shoreline from Duyen
zones and when it has direction perpendicular
Hai, Thanh Phong and Ba Tri districts.
to northeast and southeast wind. Most of
 T.T. Van, T.T. Binh I VNU Journal of Science, Earth Sciences 25 (2009) 217-222 221

. ,,...,.,,.,;,., ... ,.,,, N

'

!
~~.,, ...... 1001.iro<
•
"
•"""'"'°"''""ltl·IOOl

•..
""
--
·~

~"""""'"''""

"" '
"

"" .
•"

"" •
"

'" "'
Fig. 3. Shoreline change in 1989-2004.

Fig. 4. Shore sections in coastal communes with Fig. 5. Shore sections in coastal communes with the
the strongest erosion in 1989-2004 ordered strongest accretion in 1989-2004 ordered decreasingly:
decreasingly: Thanh Hai; (b) Thua Due; (c) Phu (a) Thanh Phong, (b) An Thanh 3, (c) Trung Binh, (d)
Tan; (d) Tan Dien, Kieng Phuoc; (e) Cu Lao Dat Bao Thuan- Bao Thach, (e) Dong Hai.
dune, (f) Dan Thanh
Shoreline 1989 in yellow, 2001 in red, 2004 in green
222 T. T. Van, T.T. Binh I VNU Journal of Science, Earth Sciences 25 (2009) 217-222

References [3] G. Winarso, Budhiman, The poiential

application of remote sensing data for coastal
[!] K. Selvavinayagam, Shoreline Change study. Proc. 22nd. Asian Conference 011 Re111ote
Monitoring in Coastal India, Using Remote Sensing, Singapore, 2001.
Sensing and GIS Tools. EzineArticles.com. 29 [4] H. DeWitt, J.R. Weiwen Feng, Semi-Automated
Aug 2008, available on: construction of the Louisiana coastline digital
http://ezinearticles.com/?ShoreLine-Change- land-water Boundary using landsat TM imagery,
Monitoring-in-Coastal-India.-Using-Remote- Louisiana's Oil Spill Research and Develop1nent
Sensing-and-GIS-Tools&id~J J54397 Progra1n, Louisiana State University, Baton
[2] Binzhao, Haiquiang Guo, Yaner Yan, Qing Rouge, LA 70803 (2002).
Wang, Bo Li, A simple waterline approach for [5] A.A. Alesheikh, A. Ghorbanali, N. Nouri,
tidelands using multi-temporal satellite image: A Coastline change detection using remote
case study in the Yangtze Delta. Coastal and sensing. Int. J. E11viro11. Sci. Tech., 4 (1) (2007)
Shelf Science, 77 (2007) 134. 61.