Rapid assessment o f water eutrophic state in tien yen ha coi bay by estimating chlorophyll a concentration using geostatistical interpolations and modis data

V N U Jo u rn a l of Scicncc, E arth Scicnces 28 2012 181-189Rapid assessment of water eutrophic state in Tien Yen-Ha Coi Bay by estimating Chlorophyll-a concentration using geostatist

V N U Jo u rn a l of Scicncc, E arth Scicnces 28 (2012) 181-189

Rapid assessment of water eutrophic state in Tien Yen-Ha Coi

Bay by estimating Chlorophyll-a concentration using

geostatistical interpolations and MODIS data

Nguyen Thi Thu Ha’’*, Pham Thi Tuyet^, Mai Trong Nhuan', Nguyen Van Vuong',

^Faculty o f Geology, VNU University o f Science, 334 Nguyen Trai, Hanoi, Vietnam

^ VNƯ Sea and Island Research Center

R eceived 25 July 2012; received in revised form 9 A ugust 2012

A bstract E uữophication is a natural process o f enrichm ent o f water by nutrients, how ever in

many case is considered as one of the major factors of water pollution affecting coastal ecosystem

Selection and application o f suitable indicator and m ethod to assess sea euừ ophication is utm ost

im portant This study aim s at clarifying w ater euứophic state m T ien Y en - H a Coi B ay using

geostatistical methods o f 40 in situ sites data and auxiliary information from MODIS/Terra image

Resultant predicted chlorophyll-a concenừation distribution map in Tien Yen - Ha Coi Bay

produced by cokriging and euừ ophic possibility m ap com puted by indicator kriging identified clearly that the b ay is classified as natural euừophic waters B y this study, M O D IS offers the

possibility to support auxiliary information for assessment euữophic state m large water areas

Cokrigm g and indicator kriging was proved to be effective for rapid environm ental quality assessm ent and coastal w ater m anagem ent

Keyw ords: Eutrophications, cokriging, indicator kriging, estim ation map, probability map.

1 Introduction

Eutrophication o f coastal w aters has been

considered one o f the m ajor threats to the health

o f m arine ecosystem s The different processes

and effects o f coastal eutrophication are well

known and docum ented [1-4] T he m ain cause

o f euừophication in coastal w aters is nutrient

over-enrichm ent (nitrogen, phosphorus and

silica) leading to: an increase in the frequency

o f harm ful algal bloom s, shellfish

contam ination, anoxic and hypoxic events and

* Corresponding author Tel: 84-1222275688

E-mail: hantt_kdc@vnu.edu.vn

fish kills; a loss o f ecosystem integrity, aquaculture production, fish stocks and am enity value; and changes in biodiversity [5] Therefore, identifying the m ost efficient w ay to diagnose the euữophic state o f coastal w aters is indispensable task to protect coastal environm ent and preserve ecosystem health

H ow ever euừophication o f coastal waters has been considered for a long tim e and many related researches have been carried-out

w orldw ide, one im portant question is still to be answ ered: “H ow should prim ary production o f eutrophication be m easured or estim ated?” Such question requires thorough scientific 181

182 N T T H a e t a l / V N U Journal o f Science, Earth Sciences 2 8 (2 0 1 2 ) Ĩ 8 1 - Ì 8 9

a n a ly se s as w e ll a s c o o r d in a tio n , n e w m e th o d s

and approaches in assessm ent o f coastal water

euữophic state need be developed and

dem onstrated C hlorophyll-a (Chl-a)

concentration is an effective m easure o f the

ừophic status o f sea and land w aters, because it

is re la te d s ừ o n g ly to a q u a tic p h y to p la n k to n

a b u n d a n c e a n d b io m a s s E s tim a tin g C h l-a

concentration is one o f the m ost traditional and

significant applications o f rem ote sensing for

evaluating aquatic ecosystem s and m onitoring

eutrophication [6-10] T his study develops a

m ethod for rapid assessm ent o f w ater eutrophic

state in Tien Yen - H a Coi B ay by using

various geostatistical interpolations o f in situ

Chl-a concenfrations and auxiliary inform ation

from M O D IS/T e ư a im ag e R esultant

probability map o f w ater eutrophic state and

potential based on Chl-a concentrations o f this

study helps delim iting zones o f high and low

eufrophication Since no other data o f sim ilar

nature is available at the present time, the

obtained results w ere considered as a first baseline for coastal w ater environm ental analysis and m anagem ent o f Tien Yen - H a Coi

B ay on eutrophication assessm ent

2 M a te ria ls a n d M eth o d s

The study area

Tien Yen - H a Coi Bay IS adjacent to China

- V ietnam border and connected to the South China Sea by five channels (Fig 1) The most rem arkable feature o f the bay IS its shallowness: the sea depth generally ranges from 2 to 5 m

A nother feature is that the bay has a strong dium al tide regim e w ith m axim um tida!

am plitude o f 5 m [11] Therefore, aquatic ecosystem and environm ent o f the bay are dom inated m ainly by oceanographic factors such as tide, waves, and near-shore currents

21.67'’N

1_I I I I I I I I

Figure 1 Location o f T ien Y en - H a Coi B ay and positions o f 40 sam pling points for w ater sampling

N T T Ha et a i / V N U lo urnal o f Science, Earth Sciences 28 (2 0 1 2 ) Í S Ì - Ĩ 8 9 183

A decline in the environm ental and

ecological systems in Tien Yen - Ha Coi Bay

has been reported, including loss o f seagrass

m eadows, phytoplankton abundance and w ater

contam ination [12] One o f serious

environm ental threats on the bay w ater

environm ent is euữophication process: its state

and cause Therefore, assessing w ater eutrophic

state in Tien Yen - Ha Coi Bay IS an

indispensable task to protect the bay

environm ent and preserve ecosystem health

W aĩer sam pling and analysis

Sea w ater in Tien Yen - Ha Coi Bay was

sam pled at 40 points using a speed-board on 6

July 2010 with a Global Positioning System

(G PS) receiver to locate the points These

points, shown in Fig 1, w ere selected to

investigate the environm ental conditions all

over the bay The samples were taken at 50 cm

w ater depth using V an D om w ater sampler,

filled in 1 liter cleaned bottles at the constant

tem perature of 4 "c storage and transported to a

laboratory W ater sam pling w as can ied out

w ithin 10 hours before/after tim e o f image

acquiring on 6 July 2011 Speed-boat w as used

for fast data collection

Concentration o f Chl-a in w ater was

determ ined by standard Specfa-ophotometric

m easurem ent m ethod [13] Firstly, w ater

sam ples were filtered by a pre-w ashed 47m m

glass fiber filter and then exừ acted into 90%

acetone The Chl-a and phaeophytin

concentrations in the extract w ere determ ined

by specữophotom eter at 750 nm and 664 nm

before acidification, and 750 nm and 665 nm

after acidification D escriptive statistics o f the

resultant Chl-a concenừations o f the 40 samples

are sum m arized in Table 1

M O D IS d a ta

T h e T e ư a s p a c e c r a ft p a s s e s o v e r T ie n Y e n

- H a C o i B a y a t a b o u t 3:2 0 G M T (1 0 :2 0 local tim e ) e a c h d ay T h is tim e IS s u ita b le fo r

a c q u irin g s a te llite im a g e ry to c o m p a re w ith in situ w ater quality M O D IS /T eưa level IB

image data acquired on the same date as the

w ater sam pling (6 July 2010), w hich were calibrated at-aperture radiances for the 36 bands

a n d g e o -lo c a te d fo r W S G -8 4 N 4 8 o f th e Ư T M

sy ste m , w e re u s e d to e s tim a te C h l-a

c o n c e n tra tio n A c c o rd in g to H a a n d K o ik e

(2011) [14], it was identified that dark-object subừaction (DOS) by Chavez (1988) [15] was a suitable atm ospheric correction m ethod for the

M ODIS data o f Tien Yen - Ha Coi Bay

T h e re fo re , th is s tu d y u s e d o b ta in e d re fle c ta n c e

after DOS as data for cross-estim ation

G eostatistical m ethods Co-Kriging

C o-K riging (CK ) is a form o f kriging that involves m ultiple variables It is considered a

‘hy brid’ K riging technique (i.e non-stationary geostatistical m ethod) In fact, the m ethod is a

m ultivariate extension o f Kriging that allows inclusion o f m ore readily available and inexpensive attributes in the prediction process [16] CK uses auxiliary variables and takes into account additional coưelated inform ation betw een variables to im prove spatial prediction,

w hich requires an estim ation o f cross- variogram s W ith CK the estim ated value at an unsam pled location is a linear w eighted sum o f all o f the variables being exam ined (i.e two or more) For two variables, the m odels are:

z , { x ) = ụ , { x ) + e ,{x) (2)

184 N T T H a e t a i Ị V N U jo u r n a l o f Science, Earth Sciences 2 8 (2 0 1 2 ) 1 8 1 -189

W here ỊẦ^ and are unknow n m ean values

(constants) and and £ ị ^ are random eư ors.

Each o f these sets o f random e ư o rs m ay exhibit

autocoưelation and cross-correlation betw een

the datasets, which the procedure attempts to

model C o-K riging uses this cross-coưelation

or covariance to im prove the estim ation o f

Z ^ { x ) The covariance is calculated according

to the following formula:

w here cov is the covariance o f two random

variables Covariance is a scaled version o f

coưelation So, w hen tw o locations, Si and Sj,

are close to each other, you expect them to be

sim ilar and their covariance (a coưelation) will

be large As S i and S j get farther apart, they

becom e less similar, and their covariance

becom es zero

Indicator K riging

Indicator K riging (IK) is a non-param etric

geostatistical m ethod for estim ating the

probability o f exceeding a specific threshold

value, 2*, at a given location In indicator

Kriging, the stochastic variable, Z(u), is

ừansfom ied into an indicator variable with a

binary d istìb u tio n , as follows:

I { u - Z ị k ) ^ { (1, i f Z(u) < z,k, k = 1,

2, m @ 0, otherw ise) (4)

The expected value o f I(u; Z t), conditional

to n surrounding data, can be expressed as:

E[l{u-, z J ( « ) ) ] = P r a ỗ { Z i u ) < z , \ i n ) } = F { u - , z , \ ( n ) ) p{ u; Zi^\{n)) = \ - F { u ; I («))

(5)

( 6 )

W h e re F (u ; Zk\(n)) is th e c o n d itio n a l

c u m u la tiv e d is trib u tio n fu n c tio n o f Z (u ) <

Z ịk, an d P (u ; Z;t|(n)) is th e p r o b a b ility that

Z (u ) > Z ị k A t a n u n s a m p le d lo c a tio n , Uo,

e s tim a tio n m u s t u se in d ic a to r k rig in g and

th e in d ic a to r e s tim a to r, / (uo; Zk), a c c o rd in g

to:

; = 1

w h e re I(uj, Z 0 re p re s e n ts th e v a lu e s o f th e

in d ic a to r a t m e a s u re d lo c a tio n s Up j =

1 ,2 , ,/i, a n d X,j(Zk) is th e w e ig h tin g fa c to r o f

I(uj, ZfJ in e s tim a tin g f (uo; Z 0

Results and Discussion The atm ospheric co ư ectio n m ethods, DOS, were applied to the M OD IS 36 bands data to

obtain surface reflectance (R ự s (Ẵ ))o ĩ the Tien

Yen - Ha Coi Bay w ater surface entire Table 1 sum m arizes descriptive statistics o f obtained

Rrs(Ằ)dX tw o bands: visible green band (band 12,

at 551 nm) and blue band (band 9, at 443 nm), using M O D IS /T eưa im age at 1 km pixel size

w ith total num ber o f pixels is 392

Table i D escriptive statistics w ater sam ples datasets o f Chl-a concenừations (mg/m^) and obtained reflectance

o f bands 9 and 11 M O D IS/T eưa Im age after DOS

Item Param eter

Chl-a (m g/m ’)

^ ( 4 4 3 ) (M O D IS band 9)

/?„(551) (M O D IS ban d 12)

N T T Ha et a i / V N U Journal o f Science, Earth Sciences 2 8 (2 0 1 2 ) 1 8 1 - Ì 8 9 185

Fig 2 show s a strong relationship between

C hl-a concentrations o f w ater sam ples and ratio

o f tw o reflectances at M ODIS band 12:

R„(551), versus band 9: R^,(443) (correlation

coefficient r = 0.78) This relationship

confirm ed a high possibility o f cokriging in

prediction C hl-a concentrations from

M O D IS/T erra image data (R,,(551)/R,,(443))

o

-s

Figure 2 R elationship betw een C hl-a concentration

o f w ater sam ple and ratio o f two reflectances at

M ODIS band 12 (551 nm) versus band 9 (443 nm)

Experim ental covariance produced from

th e s e tw o d a ta s e ts w a s a p p ro x im a te d b est by

the exponential model with a nugget effect o f

0 1 2 m g / m \ a sill o f 16.6 m g / m \ a n d a ra n g e o f

17 km (Fig 3A) To check the spatial estim ation accuracy o f CK with a cross- validation, a scattergram that represents the

re la tio n s h ip b e tw e e n th e in situ ^ C h ỉa - v alu e

a n d th e p re d ic te d C K v a lu e u sin g th e

Cchia data w as produced (Fig 3B) The

re s u lta n t m e a n e ư o r o f th e p re d ic tio n is c lo se to

0 (0.005 mg/m^) and the RM SE is relatively

low (0 9 9 m g /n v '), w h ic h c o n firm s th e high

c a p a b ility o f C K T h e re s u lta n t C K d istrib u tio n

o f Chl-a concentration and the kriging variance map are depicted in Fig 4 and 5 The kriging variances are negligible over the entire bay

Only a sm all zone outside the bay, where no in situ data w as collected, has a relatively high

variance, but its value is under 10% o f the

estim ated (^Chia Such variance trend also

supports the correctness o f the CK estimation

Figure 3 (A ) C ovariance and exponential m odel (curve) o f the in 5/7uChl-a concentrations andR „(551)/R „(443) ( B ) Scattergram for cross-validation o f cokriging prediction The 45-degree line is superim posed

186 N T T H a et al. / V N U Journal o f Science, Earth Sciences 2 8 (2 0 1 2 ) Ĩ 8 Ĩ - 1 8 9

Prediction Chl-adừtribution map produced by CK

The spatial distribution o f Chl-a

concenứations w ithin Tien Y en - H a Coi Bay

was predicted by C K interpolation (Fig 4)

Basing on the map, Chl-a concentrations on 6

July 2010 ranged from 8.1 to 1 6.5 m g/m ^

w hich are 4 to 7.5 tim es o f the eutrophic level,

2.21 mg/m^of Sim boura et al [17] Generally,

concentrations o f Chl-a over 14.3 mg/m^ occur

in local estuaries at the m ouths o f the Ha Coi and D am Ha Rivers and along the coast from

M ong Cai to H ai Ha The m iddle bay also contains high C hl-a from 13.8 to 14.3 mg/m^ from Cai Chien to Hai Ha districts On the confrary, the concentrations becom e low tow ard the w est near Cai B au Island w ith a m inim um

o f 8.1mg/m^ in the channels connected w ith the outer sea

21.55»N

107.45‘F.

ÍOS.DO-E

C h l-a (n ig /m ^

■ i 15.15- 16.50 14.32- 15.15

H i

H i ) 3 4 9 - 13.81 13.29- 13.49 12.97- 13,29

i S 12.45- 12.97 11.62- 12.45

1 0 2 7 -1 1 6 2

8 1 0 - 10.27

21.15-’N

Figure 4 Spatial distribution of Chl-a concenứations produced by cokriging

21.55-'N I08.00“E

V ariance o f

C h l-a (mg/in')^

WÊÊ

■ 1 l.2 t

H I 1.00

I Ẽ M Ổ.83

0 7 0 -O.s:^

0 5 8 - 0 7 0

0 4 9 - 0 5 8 0.38 - 0.49

1.77

1.46

1.21 1.00

107.45«E I _ £

21.15‘N

Figure 5 Kriging variance for representing uncertainty o f estimation o f Chl-a concenữations by cokriging

N T T H a et al. / V N U jo u r n al o f Science, Earth Sciences 2 8 ( 2 0 Ì 2 ) 1 8 Ĩ - Ĩ 8 9 187

Those spatial characteristics o f Chl-a

concentrations conform ed to the hydrodynam ic

system in the bay that is generated by the

interaction o f regional surface cuư ents, tides

and waves The local river estuaries and central

bay are shielded by the islands (V an N uoc, Cai

Chien, and V inh Thuc) and the tortuous

coastline, and therefore the hydro-energy

related to currents and w aves is w eak there

U nder such conditions, phytoplankton, the main

source o f high Chl-a concentrations, accum ulate

and grow In contrast, in the connection

channels and the largely m ovable w aters

affected directly by the outer sea, the hydro­

energy is strong These w aters contain abundant

am ounts o f re-suspended material, which prevent the accum ulation and growth o f phytoplankton Thus, Chl-a concentrations becom e low in such high-energy zones

Water eutrophic state probability map

CSTT[18] defined 10 mg/m^ o f Chl-a as the

E nvironm ental Q uality Standard (EQS) for coastal waters I f the C hl-a concenừation o f a

w ater body frequently exceeds this criterion in summer, it is regarded as having a eutrophic condition B ased on this criterion, the waters o f Tien Y en - H a Coi Bay w ere considered to be euưophic

21.55"N

107.45-E

Cai Bau

108,00*E

P{u;\ồ)

■ ■ 0.97 - 1.00

■ ■ 0.94 - 0.97

m 0 8 9 -0 9 4

B | 0 8 4 -0 8 9

ggsi 0.56- 0.68

0.42-0.56 0.23 - 0.42 0.00-0.23

Figure 6 P robability map o f w ater euừophic state and potential in T ien Y en - Ha Coi Bay by IK

Figure 6 shows probability o f w ater

euừophic state and potential in Tien Y en - Ha

Coi Bay by using IK and threshold o f 10

m g/m ^ A high probability o f euừophication

(P(u;10) ranges from 0.97 to 1) w as determ ined

in the bay w ater entire except for the zones in

front o f Tien Yen Estuary or channel connected

the bay w ith outer sea w hich located between

V an N uoc and Cai Chien islands However, undesirable eutrophication disturbances, such as algae bloom s, generated in the w ater with concentration o f C hi-a over lOOmg/m^ [19] have not been observed in Tien Yen - Ha Coi Bay This bay m ay be eutrophic by natural

188 N T T H a et al. / V N U jo u r n a l o f Science, Earth Sciences 2 8 (2 0 1 2 ) 181-189

processes rather than by anthropogenic causes

such as waste water, thereby m aintaining a high

ecological quality sim ilar to other coastal areas

[20].

Conclusions

This study dem onsfrates usefulness o f two

geostatistical m ethods, CK and IK, and

auxiliary inform ation for coastal w ater

euừophication assessm ent by considering the

relationship betw een in situ data and M OD IS

image reflectance, analyzing the excess o f Chl-

a concentration against the environm ental

standard A case study for Tien Y en - Ha Coi

Bay clarified that Chl-a concentrations in

seawater exceeded high eufrophic level [17] and

the bay can be m arked as natural eufrophic

waters

A nother superiority o f the com bination

betw een geostaitistical m ethods and rem ote

sensing data was to provide rapidly database for

good understanding o f the spatial variation and

distribution o f Chl-a concenừation

Consequently, that is easy to delineate high and

low risk areas o f eufrophic state and potential

for regional w ater m anagem ent

Acknowledgement

W e are grateful to Projects coded Q GTD

10.31 TRIG -A and 105.09.82.09 N A FO STED

by the V ietnam ese N ational Scientific G rant for

their support o f the fieldw ork and sam ple

analysis Thanks also to N A SA for providing

the M ODIS data

References

[1] Anderson, D M., Gilbert p M., and Burkholder

Euừophication: Nutrient Sourccs, Composition,

and Consequences.” Estuaries 25 (4b) (2002)

704^726

[2] Diaz, R J., Nestlerode J., and Diaz M L., 2004

“A global perspective on the effects of eufrophication and hypoxia on aquatic biota” InG L Rupp and M D White, eds Proceedings o f the 7th International Symposium

on Fish Physiology, Toxicology, and Water Quality, Tallinn, Estonia, May 12-15, 2003

Division (EPA 600/R-04/049)

[3] Lu, S and Hodgkiss I J., “Harmful algal bloom causative collected from Hong Kong waters.”

N ydrobioỉogia.5l2(ì-3) (2004) 231-238

[4] Spokes, L J and JickcHs T D., “Is the atmosphere really an important source of

ContinentalShelf Research 25 (2005) 2022-

2035

[5] Shafique N A., Fulk F A., Autrcy B c , and

Sensing o f Water Quality Parameters for Large Rivers in the Ohio River Basin, Ohio River Basin Consortium for Research and Education, Marietta, OH, November 5-7, (2003) pp 216 -

2 2 1 [6] Matthews A M., Duncan A G., Davison R G.,

An assessment o f validation techniques for estimating chlorophyil-a concentration from

airborne multispectral imagery International Journal o f Remote Sensing, 22(2001) 429-447.

[7] Cauwer V D., Ruddick K., Park Y J., Nechad B., Kyramarios M., Optical remote sensing in support o f euừophication Monitoring in the

southern North Sea EARSeL eProceedingSy

3(2004)208-221

[8] Zimba p V and Gitelson A., 2006 Remote estimation o f chlorophyll concentration in hyper-eutrophic aquatic systems: Model tuning

and accuracy optimization Aquaculture, 256;

272-286

chlorophyll-a concentrations in coastal waters

with varying suspended matter and CDOM

concentrations In: Richardson L L and LeDew

N T T Ha et al. / VhJU Journal o f Science, Earth Sciences 2 8 (2 0 1 2 ) Ĩ 8 1 - Ì 8 9 189

E.F (Eds.), Remote sensing o f Aquatic Coastal

F.cosystem Processes: Science and Management

Application, spnnger, Netherlands, (2009) 27-

78

[10] Ibrahim A N, Mabuchi Y., Murakami M.,

Remote sensing aigorilhms for monitoring

Hydrological Sciences Journal, 50 (2009) 525-

542

[11] NOAA, lUCN, VASI, Khuôn khổ quàn lý tổng

hợp vùng bờ Quảng Ninh - Hải Phòng Báo cáo

tổng kết Dự án Hà Nội, 2009, 44p

Programme), 2004 Scagrass in the South China

Sea Technical document o f Project: “Reversing

Environmental Degradation Trends in the South

China Sea and G ulf of Thailand’’, UNEP,

Bangkok

Examination o f Water and Wastewater, 20th ed

Washington DC

[14] Ha N T T., Koike K., Integrating satellite

imagery and geostatistics o f point samples for

monitoring spatio-temporal changes o f total

suspended solids in bay waters: application to

Tien Yen - Ha Coi Bay (Northern Vietnam)

Frontiers o f Earth Science, 5 (201 1) 305-316.

[15] rh a v ez J p s Jr., An improved dark-object subtraction technique for atmospheric scattering

Sensing o f Environment, 24 (1988) 459-479 [16] McBratney A.B., Odch I.O.A., Bishop T.F.A., Dunbar M.S and Shatar T.M., An overview of pedometric techniques for use in soil survey

Geoderma, 97 (2000) 293-327.

[17] Simboura N., Panayotidis p., Papathanassiou E.,

A synthesis o f the biological quality elements for the implementation of the European Water Framework Directive in the Mediteưanean ecoregion: The case o f Saronikos Gulf

Ecological Indicators, 5 (2005) 253-266.

[18] CSTT, 1997 Comprehensive studies for the purposes o f Article 6 & 8.5 o f DIR91/271 EEC, the Urban Waste Water Treatment Directive, 2nd edition, Ireland

[19] Cannizzaroa J p., Cardcra K L., Chena F R., Heilb c A., Vargoa G A., A novel technique for detection o f the toxic dinoflagcllate, Karcnia brevis, in the G ulf o f Mcxico from remotely sensed ocean color data Continental Shelf Research, 28 (2008) 137-158

Biodiversity in coastal zones o f Tien Yen - Dam

Ha, Quang Ninh Province and Conservation Proceeding o f the 2nd National Confcrcncc on Biodiversity, Hanoi 17 October 2010, pp: 61- 74