2012 status and threats on seagrass beds using gis in vietnam

Seagrasses, marine flowering plants, are widely distributed along temperate and tropical coastlines of the world. Seagrasses have key ecological roles in coastal ecosystems and can form extensive meadows supporting high biodiversity. Till now, fourteen seagrass species belonging to four families were found in Vietnam: Halophila beccarii, H. decipiens, H. ovalis, H. minor, Thalassia hemprichii, Enhalus acoroides, Ruppia maritima, Halodule pinifolia, H. uninervis, Syringodium isoetifolium, Cymadocea rotundata, C. serrulata, Zostera japonica and Thalassodendron ciliatum. A total area of seagrass beds in Vietnam is estimated to be approximately 17000 ha by satellite images and GIS technology. In recent years, the distribution areas and densities of seagrass beds in Vietnam have been serious decreased compared with those 10 – 15 years ago. The decline level depended on the impacts by the natural process, the economical activities and the conservation awareness of local people. Thus, it is different at each coastal area. Generally speaking, the distribution areas and densities of seagrass beds were decreased by more than 50%. Seagrasses on tidal flats in some areas such as Quang Ninh, Hai Phong, Phu Quoc seem to be nearly lost. The distribution areas of seagrass beds in 2009 at Tam Giang – Cau Hai lagoon and Cua Dai estuary was decreased by 50 – 70% of those in early 1990s.

Status and threats on seagrass beds using GIS in Vietnam

Cao Van Luong*a, Nguyen Van Thaoa, Teruhisa Komatsub,c, Nguyen Dac Vea, Dam Duc Tiena

aInstitute of Marine Environmentand Resources, Vietnam Academy of Science and Technology, 246

Da Nang, Ngo Quyen, Hai Phong, Vietnam; bAtmosphere and Ocean Research Institute, The University of Tokyo, 5-1-5, Kashiwanoha, Kashaiwa, 277-8564, Japan; cJapan Science and Technology Agency, CREST, 4-1-8 Honcho, Kawaguchi, Saitama, 332-0012, Japan

ABSTRACT

Seagrasses, marine flowering plants, are widely distributed along temperate and tropical coastlines of the world Seagrasses have key ecological roles in coastal ecosystems and can form extensive meadows supporting high

biodiversity Till now, fourteen seagrass species belonging to four families were found in Vietnam: Halophila beccarii,

H decipiens, H ovalis, H minor, Thalassia hemprichii, Enhalus acoroides, Ruppia maritima, Halodule pinifolia, H uninervis, Syringodium isoetifolium, Cymadocea rotundata, C serrulata and Thalassodendron ciliatum A total area of

seagrass beds in Vietnam is estimated to be approximately 17000 ha by satellite images and GIS technology In recent years, the distribution areas and densities of seagrass beds in Vietnam have been serious decreased compared with those

10 – 15 years ago The decline level depended on the impacts by the natural process, the economical activities and the conservation awareness of local people Thus, it is different at each coastal area Generally speaking, the distribution areas and densities of seagrass beds were decreased by more than 50% Seagrasses on tidal flats in some areas such as Quang Ninh, Hai Phong, Phu Quoc seem to be nearly lost The distribution areas of seagrass beds in 2009 at Tam Giang – Cau Hai lagoon and Cua Dai estuary was decreased by 50 – 70% of those in early 1990s

Keywords: seagrass, Vietnam, GIS, satellite image

1 INTRODUCTION

Ecosystems such as mangroves, seagrasses and coral reefs play important roles in sustaining resources and healthy coastal environment These ecosystems have high productivity also providing valuable coastal habitats for animals and also epiphytic flora From the ecological point of view, they create proper environments as ecosystem engineers They stabilize bottom sediments [1, 2], maintain coastal water quality and clarity, buffer water flow [3] Additional effects of seagrass beds seem to be similar to those of seaweed forests such as pH distribution [4], dissolved oxygen distribution [5], light distribution [6] and water temperature [7, 8, 9] These functions may be very important for sound coastal ecosystem of seagrass Thus, spatial distributions of seagrass beds are an important factor for realizing sustainable development of coastal waters However, seagrass beds are threatened by human activities such as reclamation, trawl, pollution and so on [10] Since it is needed to identify changes in the distribution areas, it is indispensable to map spatial distributions of seagrass beds

Because seagrass beds are distributed over large areas and under water (seagrass and coral ecosystems), they should be survey in the sea Diving and observation from the boat are traditionally used for mapping bottom habitats However, these methods are laborious and time consuming [11, 12] These surveying methods are expensive and difficult for areas without logistic supports Thus, it is desired to develop efficient mapping and monitoring systems of coastal areas Remote sensing data provide an overview over a large area They are shot regularly to meet the demand of users Since the 1970s, numerous studies have been conducted to develop methods of mapping the distribution of seagrass and coral beds using remote sensing data combined with ground survey Remote sensing data have been used for mapping seagrass and coral distributions since 1975 just after Landsat MSS data have been available Since then, many researches have been conducted and developed satellite image processing methods for mapping seagrass distribution such as Ackleson,

1987 [13], Armstrong, 1993 [14]; Congalton, 1999 [15], Sagawa et al., [16, 17] These studies were focused on waters with high transparancy or very shallow waters On the other hand, in turbid waters such as lagoons, classification

Remote Sensing of the Marine Environment II, edited by Robert J Frouin, Naoto Ebuchi, Delu Pan, Toshiro Saino, Proc of SPIE Vol 8525, 852512

© 2012 SPIE · CCC code: 0277-786/12/$18 · doi: 10.1117/12.977277

Proc of SPIE Vol 8525 852512-1

IW IIl

accuracies using remote sensing datat are significantly reduced Today, with ground surveys and support of spectrometer and ground surveys, algorithms processing remote sensing image for mapping the distribution of seagrass and coral beds are increasingly innovated and become more precise However, the use of remote sensing data for mapping coral and seagrass distributions in Vietnam is still limited up to now

In this study, the AVNIR – I, ALOS AVNIR – II, Landsat TM and SPOT 5 satellite image data have been combined together actual survey data is processed to map seagrass distribution and assess changes in their area

2 MATERIAL AND METHOD

2.1 Material and methods

We analyzed 32 satellite images: AVNIR - I in 1997, ALOS AVNIR - II in 2007 – 2010, Landsat TM in 1990 and 2000, SPOT 5 in 2009 and Topographic maps by Defense Mapping Agency of VietNam in 2002 with Coordinate system

VN-2000 (Figure 1)

Figure 1 Mozaic of satellite images covering all coastline in Vietnam

This study consits of many researches domestically funded, a number of bilaterally-funded projects relevant to seagrass management implemented in Vietnamese coastal areas (see Table 1) The results of the field surveys have been used to classify bottom substrates when satellite images have been analyzed

Proc of SPIE Vol 8525 852512-2

Table 1 Summary of past and ongoing projects related to seagrass in Vietnam

1986 - 1989 Rational utilization of typical ecosystems: coral reefs, estuaries and lagoons - code KT.03.11 Coastal zones of Quang Ninh, Hai Phong, Nam Dinh, Thua Thien - Hue and Ba Ria - Vung Tau provinces

1996 Wise management of the Tam Giang-Cau Hai lagoon, Thua Thien Hue province – code KT.DL.04-09 Thua Thien Hue coastal zone lagoons (including seagrass component)

1997 - 1999 Surveys of species composition and ecology of seagrasses in the Vietnam coastal zone Coastal zone and some islands of Vietnam

1999 - 2000

Survey of seed grounds, spawning grounds of economic aquatic species of Thua Thien Hue

1999 - 2000 Prediction of resilience and recovery of disturbed coastal communities in the tropics (SE Asia),

PREDICT

Gia Luan tidal flat, Cat Ba island

2001 Initial study on transplantation of seagrasses in Ha Long Bay Ha Long Bay

2002 - 2005 Investigation of measures for protecting and restoring disturbed seagrass and coral ecosystems Nha Trang bay, Cat Ba, Ha Long bay

2004 - 2006

Assessment of seagrass resources in coastal areas of Central and South West of

Vietnam proposedsolutions sustainable use

of resources

Coastal zone of Thua Thien Hue, Da Nang, Quang Nam, Quang Ngai, Phu Yen, Khanh Hoa, Binh Thuan,

Ba Ria – Vung Tau and Kien Giang provinces

2008 - 2010

Vietnam and proposed solutions to the sustainable management

Coastal zone of Nghe An, Thua Thien Hue, Quang Nam provinces

2009 - 2011 Study basis scientific, legal to the evaluation and compensation for oil pollution damage caused by in

Vietnamese sea area

Coastal zone of Thua Thien Hue, Quang Nam, Ba Ria – Vung Tau province

2009 - 2011

Construction overview of biodiversity status and using of seagrass ecosystems in coastal Vietnam in

2.2 Method

Geometric correction

Geometric correction of satellite images is an important step in digital image processing because spatial accuracy of the results depends on the geometric correction Satellite images have been often distorted and shown geometric deviation, caused by technical problems of endogenous and exogenous deviations such as changes in angle from the center to the edge of the image on coastal area scenes, changes in a height of the satellite above the ground due to topographical factors, the rotation of the earth and so on Calibration is the process of assigning geometric coordinates of the map to image coordinates, and then to project sampled picture to the coordinates of the map with a grid of squares on a plane

We applied geometric correction of image distortions according to the procedure mentioned above

Geometric correction is done by specifying the control points on the original image and on the reference map, such as terrain The accuracy of the control points is shown by calculating the square root of the average deviation (RMS-error)

Proc of SPIE Vol 8525 852512-3

Concerning typical image correction with high accuracy, the RMS-error is less than 1 under the uniform distribution of the control points

The UTM maps with a scale of 1 : 50000 was established in Vietnam in 1965 from airplane photos, additional revised and reprinted in 1998 and 1999 They were used for finding control points for geometric correction of the images We selected control points with little changes in surface topography such as intersection of highways and those of rocks, bridges and culverts spreading over the whole image The image coordinate system was corrected and adjusted to VN

2000

RMS-error of the geometric correction of satellite images were within 0.53 To ensure the results of the geometric correction with high precision, the total RMS-error must be less than or equal to 1 pixel size of the image Our results agreed with this standard that is less than 1 Since satellite images have often covered a large area of ea surface, it is impossible to select control points on the sea surface In this case, the geometrical correction can not be obtained about some images To overcome the problem, road class information of UTM map was superimposed over the image after editing to assess the accuracy

Radiometric correction

Process converting DN values of an image to physical ones permits us to separate an object from the others on the ground Selection of image bands and their combination is very important for identifying differences of ground objects Color combinations make us to observe objects more clearly as well as combinations with those and or other channels for separating objects with higher precision such as combination of spectral channels from PCA component analysis and NDVI vegetation index Two types of color combinations were used for satellite image processing: combination of true color R:G:B allocating bands 1:2:3 (see Figure 2) and pseudo color R:G:B allocating bands 4:3:2 (see Figure 3) SPOT 4 can not be used in these combinations because blue band was not available Thus, we used pseudo color combination (R:G:B = 3:2:1) meaning three bands consisting of near-infrared, red and green, respectively

Figure 2 True color image on Tam Giang lagoon belonging to Tam Giang – Cau Hai lagoon system

Proc of SPIE Vol 8525 852512-4

Figure 3 Pseudo color image on Tam Giang lagoon belonging to Tam Giang – Cau Hai lagoon system

Atmospheric correction

Light reflected from the ground to the sensor when passing through the atmosphere is changed in light intensity due to scattering, absorption or diffusion effects with atmospheric molecules or aerosols Thus, light spectrum on the satellite image reflected from the ground is interferred by atmosphere There are many methods for atmospheric correction The method of Armstrong (1993) [14] is simple and easy to apply to satellite image processing while ensuring technical requirements This method is based on subtraction of dark pixel from other pixels A large number of pixels are taken from deep water where the influence of the bottom is negligible Thier mean value of each spectral channel is calculated and subtracted from values of other pixels

The satellite image processing softwares such as PCI and ENVI comprise atmospheric correction program They need additional inflormation on image such as receiving time and date of image and the center coordinates of the image for use of satellite image processing We applied PCI 9.1 software to correct an effect of stmosphere in this study

Water column correction

Water column correction is required for processing satellite imagery to identify and map underwater habitats When light passes into the water, its intensity is decreased exponentially with increasing depth This process is known as the attenuation of light influencing on remote sensing data to detect deep bottom habitats Degree of light reduction varies with wavelength of electromagnetic radiation In visible light, a red part of the spectrums (longer wavelengths) is decreased faster than blue one (shorter wavelengths) According to increase in water depth, information of spectrums is decreased Consequently, we can not separate bottom habitats based on thier characteristics of spectrums For example, the spectrums of sandy bottom and other habitats at 2 m deep comprise more inflormation than those at 20 m deep Spectral information of sandy bottom at 20 m in deep can be the same as that of bottom with seagrass cover at 3 m deep Therefore, the light received at the sensor depends on bottom substrates and water depth.Spectrums recorded by the sensor depend on both reflectance of bottom substrates and water column on the bottom It is necessary to eliminate the disturbance caused by the water column depth

Proc of SPIE Vol 8525 852512-5

Light intensity is exponentially decreased by increasing depth as a result of absorption and scattering Absorption process includes conversion of electromagnetic energy into other forms such as heat or chemical energy, and that by objects in water such as phytoplankton, suspended organic matter particles, dissolved organic compounds and water molecules This absorption depends on the wavelength Electromagnetic waves interact with particles suspended in water and change direction This scattering process mainly caused by organic matter and inorganic particles that increase turbidity

To eliminate the influence of water column on reflectance of substrates, it is necessary to measure bottom depth of each pixel and know characteristics of light attenuation of the water column (dissolved organic matter content) However, it is very difficult to obtain the information such as spatial distribution of bottom depth due to lack of updating bottom topographic maps and also precise horaizontal scales Lyzenga (1997) [18] offered a simple approach based on the image

to compensate for the effects of the water column This study adopted Lyzenga’s method to remove influences of water column on satellite images

Supervise classification

Satellite images combined pseudo color (3:2:1) showed that mangrove and underwater plant areas were fresh red and in dark green in color Identification among aquatic grass, seaweed and seagrass was very hard based on the satellite images because their reflectance spectrums do not differ much Then, the field survey data and supervise classification were used to classify these substrates

GIS methods

A GIS method was used to map habitat distributions of mangrove, seagrass and coral and evaluate their temporal changes in their spatial distributions However, it was impossible to detect spatial distribution of seagrass beds in turbid waters from satellite images with the supervised classification Therefore, mapping the spatial distribution of seagrass beds in such areas is based on field survey data of the projects and the results of the previous surveys (Table 1) We used AcrGIS version 9.3 as the GIS software

3 RESULT AND DISCUSSION

3.1 Status on seagrass beds in Vietnam

Figure 4 and Table 2 show obtained results on seagrass beds in Vietnam There are a total of 32 distribution maps of seagrass beds along the Vietnamese coast This study indicates that the distribution area and densities of seagrass beds in Vietnam has been seriously decreased compared with those 10 – 15 years ago Generally speaking, the distribution areas and densities of seagrass beds are decreased by above 50% We present some examples in Figures 5 and 6

The status on area of seagrass in Tam Giang – Cau hai lagoon was compared between in 1999 and in 2009 Table 3 shows that after 10 years, the are of seagrass in Tam Giang - Cau Hai remaining 1000 ha (decreased 60%)

3.2 Threats against seagrass beds in Vietnam

In Vietnam, a recent decline of seagrasses is due to their natural fluctuations and also human impacts However, researches on the natural and human impacts on seagrass ecosystems are still limited in Vietnam Threats against the seagrass ecosystems are typhoon, shipping and accidental oil spills, siltation due to soil erosion, over-fishing, aquaculture, tourism and destructive fishing methods as described in a report on vulnerability and wane of seagrasses in some key beds in Vietnam, such as: Cua Dai, Tam Giang – Cau Hai, Con Dao, Phu Qui and Thuy Trieu [19]

Proc of SPIE Vol 8525 852512-6

Figure 4 Distribution maps of seagrass beds along the Vietnamese coast

Proc of SPIE Vol 8525 852512-7

Table 2 Distribution localities, seagrass species, areas of seagrass in Vietnam [19, 20, 21]

Enhalus acoroides, Thalassia hemprichii, Cymodocea serrulata, Halodule uninervis, Halodule pinifolia, Halophila ovalis, Halophila

7 Cu Mong Lagoon Enhalus acoiroides, Thalassia hemprichii, Halophila ovalis, Halodule uninervis, Cymodocea rotudata 250

8 Thuy Trieu Lagoon Enhalus acoroides, Thalassia hemprichii, Halophila minor, Halophila ovalis, Halodule uninervis, Halophila beccarii, Ruppia maritima 800

10 Nha Phu Lagoon Enhalus acoroides, Thalassia hemprichii, Halophila minor, Halophila ovalis, Halodule uninervis, Halophila beccarii, Ruppia maritime 30

17 Hon Khoi Enhalus acoiroides, Thalassia hemprichii, Halophila ovalis, Halodule uninervis, Cymodocea rotudata 100

21 My Hoa – My Tuong Enhalus acoiroides, Thalassia hemprichii, Halodule uninervis, Cymodocea rotudata 15

22 My Giang – Ninh THuy Enhalus acoroides, Thalassia hemprichii, Halophila minor, Halophila ovalis, Halodule uninervis, Cymodocea rotundata 80

26 Van Phong Bay Enhalus acoroides, Thalassia hemprichii, Halophila minor, Halophila ovalis, Halodule uninervis, Cymodocea rotundata 200

Enhalus acoroides, Thalassia hemprichii, Cydomocea serrulata, Cymodocea rotundata, Halodule uninervis, Halodule pinifolia, Halophila

Proc of SPIE Vol 8525 852512-8

1).7-nr 1,27.41172 117,1p72 or;ro-c

LEGEND

set* 1750000 -& "Ira ss

F-1 S.:rdJAuO

I 1 A/jlV WNJM PMdS Wnd

-FUaa

7r

Figure 5 Map showing status of seagrass in Tam Giang – Cau Hai lagoon in 1999

Figure 6 Map showing status of seagrass in Tam Giang – Cau Hai lagoon in 2009

Proc of SPIE Vol 8525 852512-9

Table 3 Comparison of spatial distributions (ha) of seagrass and aquatic grass in brackish water in Tam Giang – Cau Hai lagoon between 1999 and 2009

Seagrass 2450,2 1000

3.3 Typhoon

Seagrasses growing in coastal waters often suffer disturbances induced by storm that are considered to be major factor in the dynamics of seagrass meadows The typhoons with huge waves and strong currents may uproot seagrasses and erode the sediment Besides the loss of seagrass, storms also cause sediment redistribution, which results in burying meadows

In 1997, Linda typhoon brought decrease in seagrass beds in Con Dao Island Although some increases in seagrasses can

be found after the typhoon, in general the decrease has been continued After the Linda typhoon, seagrass species

population (Thalassodendron ciliatum) had completely disappeared Linda typhoon in 1997 has direct negative effects

on seagrasses because the sediment is re-accumulated during the typhoon The Linda typhoon caused the loss of approximately 20-30% seagrass area in Con Dao Isalands [22]

3.4 Turbidity

Increase in water turbidity explains loss of seagrasses Although turbidity of sea water is influenced by the tidal and seasonal cycles, it demonstrates a trend following monsoon exposure The run-off from the rivers plays an important role for the sedimentation in the rainy season, and particularly makes increase in turbidity of water column Furthermore, the increased turbidity also occurs due to constructive activities in the land [23]

3.5 Destructive fishing methods

Destructive fishing practices impact seriously to seagrass beds They include fishing with explosives, trawling, gill net and using chemical cyanide etc Trammeling and gleaning animals and plants on intertidal flats affect seagrass beds Digging the bottom for taking molluscs and other marine organisms by coastal dwellers caused also the seagrass loss Dredging the sea bottom with seagrasses for construction of canals results in disturbance of bottom sediments and increase in turbidity, which directly affect seagrass meadows Dredging brings erosion and siltation of the coasts

3.6 Economic development

The coasts are major economic development zones for the ports, trades and tourisms Rapid economic growth leads coastal environment and habitats to deterioration [24] For instance, infrastructural building activities in Con Dao after typhoon deteriorated seagrass habitats with pressures of urbanization The results of annual monitoring from 1998 till

2000 on seagrass cover, density and biomass parameters in Con Dao showed that the recovery of seagrass beds was slow, whilst some human activities, such as development of building infrastructures, quantity of fishing boats and service activities etc have been increased rapidly [23] All these activities caused environment degradation that brought the loss

of seagrasses

Proc of SPIE Vol 8525 852512-10