Mapping land cover change of the coastal area in kien thuy and do son districts hai phong province from 2001 2016

MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT VIETNAM NATIONAL UNIVERSITY OF FORESTRY STUEDENT THESIS MAPPING LAND COVER CHANGE OF THE COASTAL AREA IN KIEN THUY AND DO SON DISTRICTS,

MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT

VIETNAM NATIONAL UNIVERSITY OF FORESTRY

STUEDENT THESIS

MAPPING LAND COVER CHANGE OF THE COASTAL AREA

IN KIEN THUY AND DO SON DISTRICTS, HAI PHONG PROVINCE

FROM 2001 - 2016

Major: Natural Resources Management

Code: D850101

Faculty: Forest Resources and Environmental Management

Student: Ho Thi Nhu Quynh Student ID: 1253090028

Class: K57 Natural Resources Management Course: 2012 – 2016

Advanced Education Program Developed in collaboration with Colorado State University, USA

Supervisor: Assoc Prof Dr Tran Quang Bao

ACKNOWLEDGEMENTS

With the consent of Vietnam National University of Forestry belonged to Ministry of

Agriculture and Rural Development faculty, I perform the study: “Mapping land cover change of the coastal area in Kien Thuy and Do Son Districts, Hai Phong Province from

2001 - 2016”

With this study, I are extremely grateful for the guidance, advice and the support of many people First, I would like to thank most sincerely and deeply to my mentor – Assoc Prof Dr Tran Quang Bao, who gave helpful advices and strong supports during the implementation and completion of this study Secondly, I also would like to give a big thank

to Dr Nguyen Hai Hoa assisted the thesis to collect data and supported for me during fieldwork time

The study could not be finished and achieved result without the enthusiastic assistance, friendliness, and hospitality of the local government and residents of two provinces Dai Hop and Bang La, I would like give a big thanks and extreme appreciation to them

I also would like to thanks to the teachers of Forest Resources and Environment Management Faculty, my friends and family who always supported and encouraged me to perform and complete the study

Because of the limited study duration as well as lacking awareness and knowledgewe are looking forward to receiving the comments, evaluation and feedback of teachers and friends to raise the quality of study and improve not only the professional knowledge but also the lacking skills of me in this study

I sincerely thank you!

ABSTRACT

Mangrove forests are among the most important and productive of ecosystems, provide habitat for wildlife and play an important role in coastal zones, which appear in the inter-tidal zones along the coast in the tropical and semi-tropical regions, (Wolandski, Brinson

et al, 2009; Tuan, Oanh et al, 2002) Monitoring the change of land cover, the method to identify which area covert mangrove forest to other land covers, is the more increasingly wide application in health of mangrove investigation The main objective of the thesis was mapping land cover change of coastal mangrove in study site during the period of 2001-2007 and 2007-

2016 Compared two techniques, which are Supervised Classification and Normalized Different Vegetation Index (NDVI), get the higher accuracy for mapping and monitoring land cover change in study area NDVI has the higher precision, around 87.6% (81.6% for Supervised Classification), which established land cover map with four classes: mangrove, wetland, water and others (grasses, dykes, road, field and non-identified) The 15-year survey showed an upward trend of the percentage of mangrove area, approximately 380ha (increased 332%) while the proportion of water and wetland area slighly declines approximately 133ha and 179ha, respectively (declined around 43% and 46%, respectively) There are a set of causes, in which one reason of this change is government focuses on mangroves protecting activities and propaganda activities So, mangroves area and quality increases continuously and mangrove allocation policy for local people to manage, protect and obtain benefits from these resources, which increase coastal mangrove forest area but also Final reason is international cooperation with developing countries such as Japan The main solution discussed in this study is efficient participating of the officials in management and protection

KEY WORDS

TABLE OF CONTENTS

ACKNOWLEDGEMENTS

ABSTRACT

TABLE OF CONTENTS

ACRNOYMS

LIST OF TABLES

LIST OF FIGURES

INTRODUCTION 1

OBJECTIVES, STUDY AREA AND METHODOLOGY 4

1 OBJECTIVES 4

2 STUDY AREA 4

3 METHODOLOGY 5

3.1 Data Sources 5

3.2 Field survey method 7

3.4 Monitoring the land cover change of coastal area of study area during the period from 2010 to 2016 11

3.6 Sociological investigation method 13

RESULTS AND DISCUSSIONS 14

1 SPATIAL STRUCTURE AND DISTRIBUTION OF COASTAL MANGROVES 14 1.2 Spatial mangrove structure in Dai Hop commune, Kien Thuy district 15

1.3 Spatial mangrove structure in Bang La commune, Do Son district 16

2 MAPPING LAND COVER CHANGE 17

4 QUANTIFICATION OF LAND COVER CHANGE DURING THE PERIOD OF 2001-2016 244.1 Monitoring land cover change 24

5 MAIN KEY DRIVERS OF LAND COVER CHANGE IN HAI PHONG DURING THE PERIOD 2001-2016 AND SOME OF POSSIBLE SOLUTIONS 30CONCLUSION 33LIMITATIONS 34

REFERENCES

APPENDIX

ACRNOYMS

LIST OF TABLES

Table 4.1: Description of land cover classification used 18

Table 4.2: NDVI’s values of object-based classification in 2016 20

Table 4.3: ASSESSING the accuracy of NDVI in 2016 22

Table 4.4: Assessing the accuracy of Supervised Classification in 2016 22

Table 4.5: Coastal mangrove area at Hai Phong city over time 24

Table 4.6: Changing land cover area of coastal area in Hai Phong from 2001 to 2016 25

LIST OF FIGURES

Figure 1.1: : Study area: (a) Vietnam map, (b) Hai Phong province, (c) Kien Thuy and

Do Son district and coastal mangrove distribution 4

Figure 3.1: Size of sample plot establishment 7

Figure 3.2: Flow chart of methodology for image classification and change mappingUsing Supervised Classification and NDVI mapping land cover change in study area in 2016 In addition, we will take the change detection, which has the highest accuracy to monitor the land cover change over time 9

Figure 4.1: Spatial mangrove distribution in Hai Phong 15

Figure 4.2: Land cover map of Hai Phong coastal area with two techniques: NDVI and Supervised Classification 21

Figure 4.3: Object-based classification in 2016 23

Figure 4.4.1: Land cover map coastal mangroves (a) Land cover in 2001, (b) Land cover in 2007, (c) Land cover in 2016 25

Figure 4.4.2: Mangrove change in Hai Phong during the period of 2001-2007 27

Figure 4.4.3: Mangrove change in Hai Phong during the period of 2007-2016 27

Figure 4.4.4: Mangrove change in Hai Phong during the period of 2001-2016 28

Figure 4.4.5: Chart of land cover changes during 15 year from 2001 to 2016 28

Figure 4.4.6: Land cover change in 2007 and 2016 compared to area in 2001 29

al 2002) – and remaining mangrove forest are under immense pressure from clearcutting, encroachment, hydrological alterations, chemical spills, storms and climate change (Blasco et

al, 2001) According to the survey of MA, 2005, from 1980 to 2013, the mangrove area of the world was lost 20-35%, Vietnam also lost 80% of the mangrove area Lack of meaningful information about plant health, water content, environmental stress, and other important characteristics are the reason of forest degradation The policy of management and protection mangrove resources has also many lacks of linkage among sectored and studies have not brought specific solutions and practices Such an efficient rehabilitation of situation requires supported by changing mangrove ecosystem research and knowledge (Walters el at, 2008)

Hai Phong is one of the coastal provinces in the East-North of Vietnam that has a mangrove area of about 4.742 ha (in 2012) and the coastline is about 125km length With such natural conditions, Hai Phong is one of the provinces of our country that has the potential development and capacity to protect mangrove resources Doan Dinh Tam and Dinh Thanh Giang in 2010 outlined the awareness of people in coastal resource management as how effects and influences of people to benefit and development of mangroves Along with

promote people’s participation in the protection the mangrove Besides that, some pilot programs that were carried out in the local people However with advantages, it seem to having some disadvantages that are unresolved While a long coastline and the mangrove area that distribute along the coastline is very convenient for the development of marine economy, the management is very difficult in the development and protection of the coastal mangrove resources in Hai Phong

With the widen application of remote sensing technology widely in people‘s lives today, the use of that technological applications in the scientific reasearch is also widely applied Change detection and vegetable indices are the application of remote sensing and GIS technology which is a powerful tool to help human-being go deeper, discover, describe, identify, supervise and assess the natural resources problems and health in coastal mangrove Various fields have successfully applied the remote sensing technology to fully exploit their advantages such as mapping, surveying the land, forest, environmental management, census, survey and assess the forest, forest classification investigation, land use-cover change, etc For example, agricultural change in Nigeria using a combination of post classification comparison and spectral comparisons between two multispectral scanner images of Pilon et al, 1988, or Castellana et al (2007) showed a new approach to perform change detection analysis based on

a combination of supervised and unsupervised techniques Especially in recent years, the remote sensing technology has been a powerful application in the study of issues relate to mangroves such as the mapping of status, investigating the change in forest, assessment study the mangrove environment, forest classification survey For example, Nguyen Hai Hoa et al (2013) assessed spatial-temporal changes in the extent and width - change in adjacent land use, fringe mangroves in Kien Giang Province Or, Hanh Tran et al (2015) performed a study

to assess the spatio-temporal dynamics of land use-cover change in a coastal area of Ca Mau Province However the research application of remote sensing and GIS technology to sudy

about mangrove is quite less and the content is not copious, especially using change detection and vegetable indices have not quite applied in this case Moreover, the study of mapping land use-cover change of coastal mangroves in Hai Phong have not been noticed much that it has practical significance, be scientifical with the development, protection and management of mangrove resource Starting from that the practical significance and scientific, thesis built the

study: “Mapping land cover change of the coastal area in Kien Thuy and Do Son Districts,

Hai Phong Province from 2001 - 2016”

OBJECTIVES, STUDY AREA AND METHODOLOGY

1 OBJECTIVES

- Investigating spatial distribution and structure coastal mangrove in study area

- Quantification the land cover of coastal mangroves in Kien Thuy and Do Son Districts, Hai Phong City during the period from 2001 to 2016

- Identifying the key drivers of land cover change over 15 years and proposing possible solutions for sustainable management of coastal mangrove in study area

2 STUDY AREA

The study area is the coastal mangrove in two communes, including Kien Thuy and Do Son Districts, coastal mangrove mainly distribution in Hai Phong City (figure 1.1)

Figure 1.1: : Study area: (a) Vietnam map, (b) Hai Phong province, (c) Kien Thuy

and Do Son district and coastal mangrove distribution

Hai Phong in generally and Kien Thuy and Do Son Districts in particular in the coastal areas of Gulf of Tonkin, the climate here is characterized by a tropical monsoon climate,

warm and rainy The temperature range is between 20-23 degrees Celsius, the highest is 40 degrees C, and the lowest is 5 degrees C The average humidity is from 80% to 85% in the highest was 100% at July, August, and September and lowest in December and May The average annual rainfall is 1600- 1800mm The average number of sunshine hours in the year

is approximatedly1700 hours

Dai Hop Commune in Kien Thuy District has the total area is 1,097.78 ha of natural, 9,492 people (2009) with mangrove area about 450 ha The number of households is 2,675 households with the average population density over 865 people per km2 Bang La Commune, Do Son District has natural area around 4,237.29 ha and 51,417 people (2010) with per capita income in 2005 of $ 1,100 (27 million VND)

3 METHODOLOGY

3.1 Data Sources

To perform this study and improve efficiency, science and inheritance of the study, thesis used data from many sources, the study about remote sensing and GIS technology, vegetation indices, its applications in the study mangroves in general The mapping, land cover change for coastal mangroves, change detection and vegetation indices in particularly such as multiple documents, legal documents, scientific data, essays, projects, scientific research in Vietnam and abroad

Thesis collected data with the following information:

- The study, texts, documents, data from the agency, the sectored level, books, dissertations, projects relating to the land cover change to study forest mangrove in Vietnam and the world

- The researching, reports, thesis, which related to distribution and structure, land cover change of coastal mangrove, dynamic coastal mangrove in Hai Phong Province, Vietnam and the world

- The natural, economic and social conditions of study area

Landsat Data

Thesis used Landsat satellite images to monitor, investigate, and interpret the spatial distribution and structure and land cover change of coastal mangrove in study area

T ABLE 3.1: Landsat image collected in study

2001 LE71260462001128SGS00 08/05/2001 30(15) m 126/46

2007 LT51260462007121BKT02 07/06/2007 30(15)m 126/46

2016 LC81260462016114LGN00 02/06/2016 30(15) m 126/46

- Source: http://earthexplorer.usgs.gov

Landsat Image Processing

We downloaded Landsat image from Earth Explorer Image clustering channels were been collected including individual spectral channels due to needing combination and composition to easy conduct steps later

After having combinative and composited image, the study area is not only coastal mangrove area but also including other areas, thus need to cut separating the study area or the study coastal mangrove to study and analyses image by the study area boundary in the newspapers and photographs

3.2 Field survey method

Sampling method

To conduct map of distribution and assess structure of coastal mangrove, thesis established plots with dimensions 30x30m (900m2), the length of the parallel plots with contour line The sample plots were been established according to the size and the following figure:

\

Figure 3.1: Size of sample plot establishment

The length of the standard plots perpendicular to the sea and far others plot at least 100 meters The first plot was be made from the outer edge of the forest range of 20 meter

In order to select the location of plots, firstly thesis will entirely conduct reconnaissance study area and then use a GPS, the navigation device to determine the coordinates of each plot represented the typical state in study area, which is be evenly distributed throughout the area

The sampling plots established throughout the study area that helped investigating the health status of the subjects of mangroves and also helped study’s interpretation of reality and verify the accuracy of the training samples for classifying coastal mangroves later

During the investigation, thesis established 47 represented typically plots, evenly distributed throughout the state for basic subjects in the study area

With 47 sample plots, we investigated the following table form on the health status of coastal mangrove study area

Table 3.3 Table form on the health status of coastal mangrove in the study area

No Species Age

(Years)

Heightunder (m)

Heighttotal (m)

DBH (m)

Canopy diameter (m)

Canopy cover (%) Coordinate Note

1

2

Where:

Height under (m): The height will measured using calibrated meters, approximately,

from the base to the lowest living branch and joined the tree canopy

Tree height (H total) (m): The height will be measured using calibrated ruler,

approximately, from the base to the highest branches of the tree growth and joined the tree canopy

DBH (m): Diameter at breast height 1.3 m: Calculating from the base by tape to 1.3m

in struck

Canopy diameter (m): The diameter will be measured using calibrated meters, by two

directs W-E and N-S and the largest canopy of trees

Canopy Cover (%): Calculating by the tube had diameter 2cm and judged by the

visual eye through the investigating experience

Thesis will establish 47 sample plots and investigate the status of mangroves in the field through GPS to define coordinates of the points, synthesize the survey will, process the data to match and put on ArcGIS to assess mapping of the healthy status of coastal mangrove forests in the study area

3.3 Mapping land cover change

Study mapped the land cover of coastal mangrove forest in 2016 based on Landsat 8 and figure 3.2 will illustrate this process

Mapping land cover change

Supervised Classification in 2016 NDVI analysis in 2016

Using Supervised Classification and NDVI mapping land cover change in study area in

2016 In addition, we will take the change detection, which has the highest accuracy to monitor the land cover change over time

Assessment accuracy of mapping

Assessment the accuracy is the comparison degree of correctness of a map or classification with real place (Foody, 2002) There are many methods of accuracy assessment for a map or classification A confusion or error matrix, however, used and promoted the most widely, which is both describe classification accuracy and characteristics errors (Foody, 2002) Therefore, to assess the accuracy of classification classes in comparing different vegetation indices with set of data point on the field in the state forests and other objects, the study used error matrix table In this study, a total of 335-reference points were surveyed in the field to serve as validation samples for mapping After that, 102 points (more than 30% of total point) were selected randomly for accuracy assessment with three different vegetation indices

The study used Kappa coefficient, a discrete multivariate technique, representing the percentage of chance agreement (Congalton, R G; Foody Gile.M, 2002; and Foody Gile M, 1992) The Kappa coefficient is been calculated in following formula:

Where r is the number of rows in matrix, xii is the number of observation in row I and column i, xi+ and x+i area the marginal totals of row I and column I, respectively, and N is the total number of observations (Bishop et al, 1975)

The study will base on collected data and documents for mapping land cover change

in the study area After determining the accuracy of mapping, the thesis conducted the thematic mapping fluctuation the study area over time

3.4 Monitoring the land cover change of coastal area of study area during the period from 2010 to 2016

Change detection

There are two common uses of satellite images, which are mapping land cover (image classification) and land cover change (change detection) (Bark el at, 2010; Song el at, 2001) This study focused on those two uses to mapping land cover change of coastal area in Hai Phong

Supervised Classification and Unsupervised Classification are two approaching techniques (Bark et al, 2010; Singh, 1989), in which a supervised technique requires ground truth points to get training sets containing information about the spectral signatures during two dates, and another is without any additional information besides the raw images considered (Bark et al, 2010) However, unsupervised technique has also some critical limitations about accuracy of ground truth points (Bruzzon and Prieto, 2002) In this case, the study performed the higher change accuracy and efficiency is supervised classification with the most commonly maximum likelihood classification used

Vegetation indices

Vegetation indices play an important role in monitoring variations in vegetation (Bunkei et al, 2007) Vegetation interacts with solar radiation in a different way than other natural materials The measuring the different spectral variations and studying their relationship to one another can provide meaningful information about plant health, water content, environmental stress, and other important characteristics across the spectrum (Harris

wavelengths reveal specific vegetation characteristics, also known as Vegetation Indices (VIs) (Harris newspaper, 2010) It designed to enhance the vegetation signal in remotely sensed data and provide an approximate measure of live, green vegetation amount and been widely used for the phonologic monitoring, vegetation classification, vegetation parameters, etc (Huter et al,1999) Vegetation indices based upon their robustness, scientific basis, and applicability, including RVI/SI (Ratio Vegetation Index/Simple Ratio), EVI (Enhanced Vegetation Index), DVI (Difference Vegetation Index), YVI (Yellow Vegetation Index), BVI (Brown Vegetation Index) and NDVI (Normalized Difference Vegetation Index), and so on

Among VIs, the NDVI is an index the most commonly used by its ratio properties, which cancel out a large figure for the noise caused by sun angles changing, topography, clouds and shadow, and atmospheric (Brunkei, 2007) It uses 2 channels to identify areas with

no vegetation and plants because it allows an accurate representation and discrimination based

on characteristics of chlorophyll in the leaves absorb light in the visible spectrum (0.4-0.7µ), and reflects the network light on the near-infrared spectral range (0.7-1µ) (Journal of Science and Technology of Nghe An, 5/2014) NDVI is been calculated as follow formula:

NDVI = (NIR - RED) / (NIR + RED) Where, as before, NIR and RED (or VIS) are the response in the near-infrared and red (or visible) bands respectively

The difference between the near-infrared and red (or visible) reflectance is divided by their sum NDVI has a range limited to a value from -1 to +1, showing clearly the distribution

of vegetation cover in the study area It also represents the different plant groups through this value on each plant There will usually divide into four levels:

From negative value to zero is water;

From the value is less than 0.1 usually soil, rock, sand or snow;

From the value of approximately 0.2 to 0.5 are bushes, grass or dry fields;

From 0.6 to 0.9 ~ 1 as trees, plants

NDVI is widely used for the study of vegetation such as estimating crop yields, the ability to crop, convert fields In addition, NDVI directly related to parameters such as surface soil layer, the photosynthesis of plants, water, biomass calculation, and so on

After assessing land use and land cover by two techniques, the thesis used supervised image classification and NDVI to mapping and monitoring land cover change and compare which technique has the higher accuracy for mapping land cover change of study area in period 2010-2016

3.6 Sociological investigation method

After determining the location study area, conducting the sampling and sociological surveys by interviewing local government and two of Dai Hop and Bang La inhabitants is essential to provide the most accurate information With population of Dai Hop and Bang La commune is 9,491 people (4/2009) and 8,765 people (2013), respectively, we constructed survey with 30 households and 300 people sociological survey with each commune (about 30% of population) in which almost interviewed people have directly get benefit from mangrove coastal In addition, through feedback and suggestions of the local people, this study can take measures to help local government, forest managers and people improve efficiently mangrove forest resources areas of research, protection and development in parallel with natural resources exploitation, fishing products

Table 3.4: Sociological survey by interviewing

No Name Address History of

RESULTS AND DISCUSSIONS

1 SPATIAL STRUCTURE AND DISTRIBUTION OF COASTAL MANGROVES

According to research of Vu Doan Thai, Effects of mangrove on coastal sediment deposition at Bang La commune, spatial mangrove distribution is likely to represent of two

coastal populations: Kandelia obovata and Sonneratia caseolaris (L.) Engl Mix forest,

however, is not much in natural conditions at Hai Phong city Hai Phong has the 125km dike length with the natural area about 152,000ha Such as some of other coastal districts, namely Nam Dinh, Thai Binh, Quang Ninh, Hai Phong provinces is one of the local that has much patential with tidal flats and mangroves

Based on the analysis of data on the mangrove area in Hai Phong is distributed in the region 2–sub-area 1 that concentrate in 4 districts and 3 counties coastal are Kien Thuy, Thuy Nguyen, Tien Lang, Cat Hai, Do Son, Duong Kinh, Hai An Through the funding of international rehabilitation program for mangrove projects, namely the plsanting program PAM 5325, the planting mangrove program of the Red Cross, the action program restoration mangrove of ACMAMG organization (Japan) Through the field survey with transect line and the using of Landsat 8 satellite images, the spatially ecological distribution of mangrove in two provinces Dai Hop and Bang La, mangrove species compositions that are quite simply,

namely Kandelia obovata, Soneratia caseolaris, Bruguirea gymnorrhiza, Aegiceras corniculatum, Acanthus Ebracteatus that is indicated in Fig 4.1

Figure 4.1: Spatial mangrove distribution in Hai Phong

According to the result of study in spatial distribution at Bang La and Do Son commune absolutely same with above hypothesis about spatial mangrove distribution of Dr Thai

1.2 Spatial mangrove structure in Dai Hop commune, Kien Thuy district

With 12km length of study area, the area of mangrove measured by ArcMap is 92.27ha,

the result of field survey showed the represent of two species: Kandelia obovate and Sonneratia caseolaris Combination of further field survey, interview and found data sources, mangroves in

2 areas have been planted since 1999 and regeneration layer has had a lot, mostly grown closely

sea dikes but majority is Kandelia obovata species at that time of study (7/2014)

Canopy diameter: The canopy diameter in average of Sonneratia caseolaris species is largest going up 4.9m In the forest stand, some Sonnereratia caseolaris trees have diameter more than 7m The canopy diameter average of Kandelia obovata species is 1.4m, in average

Canopy cover: that shows a multipurpose ecological indicator which is useful for

distinguishing different plant, estimating functional variables, being a intermediate stage in distinguishing the signals reflected from forest canopy and forest floor, and so on (Korhonen,

2006) At Dai Hop commune, Kien Thuy district, Kandelia obovate species was planted at a

program in 1999 After field survey indicated that canopy cover of coastal mangrove was nearly 87% in study sites and stretched over 410ha from 650m to 720m width toward the sea

1.3 Spatial mangrove structure in Bang La commune, Do Son district

This study illustrated some of coastal mangrove species at Bang La commune, Do Son

district including Sonneratia caseolaris, Kandelia obovata, Bruguiera gymnorrhiza, Avicennia marina, Acanthus ebracteatus Two main species, however, with largest density in study area are Sonneratia caseolaris and Kandelia obovata, where distributed along the dike

and the same mangrove plantation program in 1999

Canopy diameter: As the result of study, the canopy diameter in average of Sonneratia caseolaris species is largest, the same figure for Dai Hop commune 4.9m In the forest stand, some trees have diameter more than 7m The canopy diameter average of Kandelia obovata species and Bruguiera gymnorrhiza species is 1.4m and 1.9m, respectively The lowest is Avicennia marina species and Acanthus ebracteatus species, 0.9m and 0.2m respectively

Canopy cover: Mangrove species at Bang La commune ranges from 90% to 95%

In here Sonneratia caseolaris species has outstanding height so it is planted near the

sea to reduce the flow rate, especially in flood conditions To get more information and easy

to following this contents, we put a table of 335 points in Appendix that have specific characteristics and coordinate of those points

In generally, spatial distribution and structure of coastal mangrove in the study is

divided into three stable parts, the common tree is Kandelia obovata, Sonneratia caseolaris, Bruguirea gymnorrhiza that has a significantly effect in protect the sea dikes, livelihood with

local people, reducing the strong velocity, limiting the sediment transports, enhancing deposition and sedimentation