VNU Ịournal of Science, Earth Sciences 25 (2009) 217-222
Application o f remote sensing for shoreline change detection in Cuu Long estuary
Tran Thi V a n 1'*, Trinh Thi Binh2
]Instiíute fo r E n vironm ent an d Resources, Vietnam N ational U niversity H o Chi M inh City 2 U niversity ofN aturcỉl Science, Vietnam N ational ư niversity H o C hi M inh City
Received 2 November 2009
A bstract. Coastal zone o f Cuu Long estuary is a place through vvhich Mekong River flows into the Eastem Sca vvith 8 estuaries. is the zone is íormed under the iníluences in the interaction b etw e en river and sea. Geologically, Coastal zone o f Mekong delta is characterised by the predominat o f clay and silt íormations. These two íactors makes the shoreline sensitive to cxogenous processcs such assedimentation and erosion.. This paper presents an application of satellite remotc sensing technology to detect and analyze the spatial changes as well as quantiíy the shoreline change in Cuu Long estuary. Landsat and Aster satellite images were used with band ratio mcthod for shorcline change detection.The results present shoreline changes maps in three pcriods: 1989, 2001 and 2004. The outcomes o f the case study can be used as an orientation for the sustainable integrated management plan o f Coastal zones.
Keywords: accretion, band ratio, erosion, change detection, shoreline.
1. In tr o d u c tio n
S horeline, the b o u n d ary b e tw e e n land and sea k eeps c h a n g in g its sh ap e an d position co n tinu ou sly d u c to d y n am ic co nd itio ns. T h e change in sh o rclin e is m a in ly asso ciated w ith w aves, tid es, vvinds, p erio d ic storm s, sea-level change, the g eo m o rp h o lo g ica l p ro cesses such as erosion an d acc retio n an d h u m an activities [1]. E rosion an d acc retio n a ffe c t h um an life, agricu kture an d a q u cu ltu re p ractice and w aterw ay tran sp o rt activ ities. D etectio n and m easu rem en t o f sh o relin e ch an g e s are an im portant ta sk in en v iro n m cn tal m o n ito rin g and
C o a s t a l zo n e m an ag em c n t. A p p ro ach es to
* C o r r c s p o n d i n g a u th o r : T e l . : 8 4 - 9 1 9 1 8 8 4 8 5 E-m ail: t r a n t h i v a n @ h c m i e r . c d u . v n
d etectin g sh o relin e ch an g es can be roughly d iv id ed into fo u r categ o ries, all o f w hich have b o th ad v an tag es an d disadvantages: (1) co n v en tio n al g ro u n d su rvey in g can achieve high a cc u racy o f m easurem ent, b u t is labor in ten siv e an d tim e co n su m in g ; (2) m odem altim etry te c h n o lo g y u ses rad ar altim eters o r la se r altim eters. It has a g reat potential, b u t the d etecto rs a re c u rre n tly less available; (3) a irb o m e im a g e ry m e asu re m en t provides su ffìcie n t p ic to ria l ỉn íb rm atio n , b u t the íreq u e n cy o f d ata acq u isitio n is low , and the p h o to g ram m etric p ro ced u re in clud ing data acq u isitio n a n d im ag e m ap p in g is co stly as well as tim e co n su m in g [2]; (4) M u ltisp ectral rem ote sen sin g sa te llite s p ro v id e dig ital im ageries in in írared sp ectral ban d s w here th e land-w ater in te ría c e is w ell d eíin ed . F u rth erm ore this
218 T.T. Van, T.T. Binh / V N U Io u n ia l o f Science, Earth Sciences 25 (2009) 217-222
m ethod has ad v an tag es: not tim e co nsum ing , inexpensive ex ecu ted co st and large ground cov erage m o n ito rin s [2,3].
S atellite op tical im ages are sim p le to in terpret an d easily o b tain ab le. A b sorption o f inírared vvavelenglh reg io n by vvater and its strong re ílectan ce by v eg etatio n an d soil m ake such im ages an ideal co m b in atio n for m appirm the spatial d istrib ution o f land an d vvater. T hese ch aracteristics o f \vater, v eg etatio n and soil m ake the use o f the im ages th at contain visible and infrared b an d s w id cly used fo r coastline m apping [4]. T h is stud y ap p lied satellite im age interpretation and G IS to d etect an d analy ze the spatial ch anges as w ell as q u a n tify th e result o f sh o relin e ch an g e in C uu L ong estuary . The results present sh o relin e ch an g e m ap s in three periods: 1989, 2001 and 2004.
2. S tu d y a r e a a n d d a t a sets 2. l.Stucty area
C oastal zo n e o f C uu L ong estu aries extends from T ranh De riv er m outh to T ieu riv e r m outh.
S tudy area is located betvveen latitude 105° 58' 58"N to 106° 51’ 8" N an d lo n g itu d e 9° 23' 55"
E to 10° 2 5 ’ 37E (F ig. 1). T h e m ainland is lim ited by b o u n d ary o f Coastal districts, including 8 d istricts: L o n g Phu (S o c T rang P rovince), C au N gang, D uyen H ai, T ra C u (T ra V inh p rovince) B inh D ai, T h an h Phu, B a Tri (B en T re p ro v in ce) an d part o f G o C o n g D ong (T ien G ian g pro vince). T h is area is relatively flat vvith a lot o f sand d u n es tu rn in g back to the sca. T h e clim ate has tro p ic and m onsoon ch aracters w ith tw o seaso ns: th e rain y from M ay to N o v em b cr and th e d ry from D ecem ber
to A pril. T h e riv e r system vvith h igh den sily is a place th ro ug h vvhich M ek on g R iv e r flow s into the E astern Sea w ith 8 estuaries.
<r^ Ệ r
2.2. D ata set
S atellite im ag es vvere ac q u ire d in thrcc dates: L an d sat T M in 1989, L an d sat E T M + in 2001 and A ster in 2004 (T ab le 1). Landsat im ages h av e sp atial resolu tion o f sp cctral bands in 30m and E T M + b an d 8 is p an ch ro m atic o f 15m reso lu tio n . A ste r im age has 2 g ro u p s o f sp ectral b an d s w ith d iffere n t pixel resolutions:
V N IR b an d s in 15m an d S W IR b a n d s in 30m.
T hese im age w c re tak en in d ry season.
T h e rc ío re sh o relin e is easily id e n tiíìed due to the fact that th e w a te r level is n o t v ery high.
T o p o g rap h ical m a p a t scalc 1:100.000, U TM co o rd in ate an d W G S -8 4 p ro ịectio n vvas use for g eo -rectify in g a n d d c te rm in in g ero sio n / accretio n locations.
T able 1. Satellite data used in th e study
A cquisition Tim e___________ Satellite_____________Sensor____________ R esoỉution
1 6 /0 1 /1 9 8 9 Landsat5 T M 30m
11/12/2001 Landsat7 ETM+ 15m,30m
12/12/2004 A ster A ster 15m ’ 30ni
1.1. VaiI, 1.1. Binh ì VNU ịournaì ọ f Science, Earth Sciences 25 (2009) 217-222 219
3 . M cth od ology
Spcclral band ra tio is o n e o f th e m ost com m on m athem alical o p eratio n s ap p lied to m ulti-spectral ìm age d ata. R atio im ages are calcu lated as the d iv isio n o f D N v alu es in one spectral band by th e co rrc sp o n d in g p ixel value in an o th cr band. B an d ratio in g p ro v id es unique info rm atio n no t a v a ila b le in an y sing le band that is u seíu l for d istin g u ĩsh in g e a rth suríace features. B and ratio o p cratio n can red u ce the en v iro n m en tally ìn d u c ed v aria tio n s in the DN v alu es o f a sin gle b an d , su c h as brightness v ariatio n s cau sed b y to p o g rap h ic slo p e and asp ect. shado w s o r scaso n al ch an g e s in sunlight illum in ation an g le a n d intensity. T h ereío re, b an d ratio in g can c m p h a siz e and highlight subtle variations in th e actual sp ectrai resp on ses o f v arious su rĩa c c covcrs.
E x p erim en ts h av e sh o w n th at green band in 0.52 - O.ÓỊ-im w a v e le n g th s (L a n d s a t b an d 2 and A ster band 1) is se n sitiv e to vvater turbidity d iffere n ces plus se d im e n t an d p o llu tio n plum es, becau se it c o v e rs th e g reen re íle c ta n c e peak from Ie a f su rfaces. It can be useful for discrim in atin g b ro a d c la sse s o f vegetation.
W ater is stro n g a b so rb c r o f n e a r in írared rad iatio n (N IR ), so L an d sat ban d 4 (0.76- O.ỌOịim) and A stcr ban d 3 (0 .7 6 -0 .8 6 n m ) are u seíu l for lo catin g a n d d c lin e a tin g w a te r bodies, d istin g u ish in g betvveen d ry a n d m o ist soil and p r o v i d i n g ì n i b r m a t i o n a b o u t C o a sta l w e t l a n d , sw am p and floo dcd arcas. L an d sat b an d 5 (1 .6 0 -1 .7 0 |im ) an d A s te r b an d 4 (1.55-1.75|i.m ) ex h ib its a stro n g c o n tra st betvveen Ịand and w a te r fe a tu re s d u e to th e h ig h d e g re e o f ab so rp tio n o f m id -in fra rc d e n e rg y b y vvater and stro n g rc ílectan ce o f m id -in frared (M IR ) by v eg etatio n and natu ral featu rcs in th is range.
T he \vaveleng th in fo rm atio n is n ece ssary in ratio in g settings. T h is stu dy u sed b a n d ratio in g m eth o d in thes m e n tio n e d w a v e le n g th reg io n to
extract soil an d w atcr from I.andsat an d A ster im ages.
Im a g e p re p ro c e ssin g : A ster im age has the sm allest spatial reso lu tio n o f 15m o f V N IR b an d s, so ban d 4 (p ix el size in 30m ) in SW IR sp ectru m w ill be d o w n sc a le d to this one. T his p ro cess is also a p p lie d to L andsat ban d s w ith 30m 1 rcsolutio n. In c h an g e detectio n , geo- rectiĩicatio n o f im ag es w ith d ifferent spatial reso lu tio n is co n stra in c d for g u aran tcein g the sm allest e rro r in o v e rla y in g rcsults. A ster 2004 w as u sed to g c o -re c tify w ith thc topographical m ap. O th er L an d sat im ages w as rcctified by A ster one. T he R M S E s w cre less than 0.5 pixel.
Im a g e a n a ly zin g : In th e íĩrst stcp, histogram th resh old m cth o d is used on N IR ban d for sep aratin g land fro m w ater. T he threshold v alu es are selcc ted so th at all w ater pixels arc sep arated from th e la n d p ix c ls . In this case, few land p ix e ls h av e b e e n m istak cn ly assig n ed to w ater pixels but no t v ice vcrsa. W ater pixels are then assig n ed to “ 1” an d land p ix els to “0” . A bin ary im ag e h as b een ach iev ed then. T his im age is n a m e d “ im a g e l” . In th e seco n d step, ra tio in g m eth o d is u sed w ith ratios green/N IR and green /M IR . G re e n /N IR ratio is u seíu l for sep aratin g lan d fro m v egetatio n, grcen/M IR for sep arating n o n -v e g e ta tio n land. W ater pixels w ill be g reater th a n 1. T w o these ratio im ages are m u ltip ly for re je c tin g m istake and íòrm ing th e n ew im ag e n a m e d “ im age2". Im ag el is m u ltip lied to im ag e2 to gen crate th ird image nam cd “ im ag e3 ". S o m e isolatcd p ix els are rem o v e w ith sie v e in g a n d filtering technique for th e íĩnal sh o relin e im ag e. T h e resu lted shoreline ex tractio n is tra n sfo rm e d in to v ecto r form at and ex p o rted in to M a p In fo form at for analyzing chang e o f e ro sio n /accrctio n areas. C han ge in pairs are e x a m in e d su ch as 1989-2001 and 20 01 -2004. Fig. 2 illu strates thc stcps o f the m ethod.
220 T.T. Van, T.T. Binh / V N U Ịournal o f Science. Earth Sciences 25 (2009) 2Ĩ7-222
S atellite im ages --- ị —
G eo m etric rectiíĩcatio n
ĩ
i
M IR b and < th re sh o ld
X
Im ag e l
ĩ
ì
G recn /N IR > 1 a n d G reen /M IR > I
Im age2
I
i
Im age3 = im ag el X im age2
Filtering . -
t
Im age4ị
r
R aster to vector
S h oreline m ap
Fig. 2. Flowchart o f extracting shoreline from satellite images.
4 . R e s u lt s a n d d is c u s s io n
A ccretion and ero sio n a re tw o inverse processes. A ccretion in creases th e area o f agricultural, forest and íìsh in g land, but yields the sedim ent in th e sea po rt, b a rin g vvatenvay transp ortation activ ities. VVhile ero sio n cau ses the loss o f land, d e stro y s co n stru c te d huildings, ho uses, th rea ten in g hum an life. T h e an aly zin g results has sh o w n th e e ro sio n /accretio n areas throug h tim e in fig. 3.
Shore is often ero d ed ro b u stly in w el!-aired zones and w hen it has d irec tio n p erp en d icu lar to northeast an d so u th east vvind. M o st o f
a ccretio n /ero sio n sectio n s in 1989-2001 are c o n tin u ed in th e nex t stage in 2 0 0 1 -2 004 vvith th e h ig h e r an d m orc rapid m easu re. Section in c o n d u e n c e b ctw ee n m id d le H au riv er and Ọ uan C hanh Bo c h a n n e l h ad ero d ed in 1989-2001, but had no c h a n g e in 2 0 0 1 -2 0 0 4 . T h is region vvas ex am in ed \vitli full v eg etatio n cov er in 3 satellite im ages. L o ts o f sectio n s vvere sed im ented in 1989-2001, b u t they w ere serio u sly ero d ed afte r 2001. It lengthened and jo in e d w ith p re v io u s ero sio n sections. It is o b v io u sly o b serv cd in sh o re lin e from D uyen H ai, T h an h P h o n g and B a T ri districts.
T.T. Van, 7.7. Binh / V N U Ịournaỉ o f Science, Eartlĩ Sciences 25 (2009) 217-222
________ựj ts ________ Sỉi________ L2________lu.
l*v»4
• I ( S M3 am c n «w
Fig. 3. Shoreline change in 1989-2004.
Fig. 4. Shore sections in Coastal communes with Fig. 5. Shore sections in Coastal communes vvith the tte stroncest erosion in 1989-2004 ordered strongest accrction in 1989-2004 ordered decreasingly:
decreasinely: Thanh Hai; (b) Thua Duc; (c) Phu (a) Thanh Phong, (b) An Thanh 3, (c) Trune Binh, (d) Tan; (d) Tan Dien, Kieng Phuoc; (e) Cu Lao Dat Bao Thuan- Bao Thach, (e) Dong Hai.
dune, (0 Dan Thanh
Shoreline 1989 in yellovv, 2001 in red, 2004 in green
222 T.T. Van, T.T. Binh / V N U Ịournal o f Science, Earth Sciences 25 (2009) 217-222
R eíeren ces
[1] K. Selvavinayagam , Shoreline Change M onitoring in C oastal India, Using Remote Sensing and GIS Tools. EzineArticles.com. 29
Aug 2008, available on:
http://ezinearticles.coĩn/?ShoreL ine-C han£e- M onitoring-in-C oastal-[ndia,-U sing-R em ote- S ensing-and-G IS -T oois& id^ 1154397
[2] Binzhao, H aiquiang G uo, Y aner Yan, Q ing Wang, Bo Li, A sim plc w aterline approach for tidclands using m ulti-tem poral satellite imagc: A case study in th e Y angtze D elta. Coastal and S helf Science, 77 (2007) 134.
[3] G . VVinarso, B udhim an, T he potcntial application o f rcm otc scn sin g data for C o a s ta l
study. Proc. 22nd. Asian ConỊerence otĩ Remote Sensing, Singapore, 2001.
[4] H. DcW itt, J.R . W eiw cn Feng, Scm i-A utom atcd construction o f thc Louisiana coastlinc digital land-w ater B oundary using landsat TM imagery, Louisiana 's Oil Spill Research and Develơpmení Program, Louisiana S tate Univcrsity, Baton Rouge, LA 70803 (2002).
[5] A.A. A lcshcikh, A . G horbanali, N. Nouri, C oastline changc dctcction using rcmotc sensing. Int. J. Environ. Sci. Tech., 4 (1) (2007) 61.