Comparing vegetation indices for mangrove forest mapping using remontely sensed data in kien thuy and do son district hai phong city

This study compared performance of the vegetation indices Normalized Difference Vegetation Index- NDVI, Simple Vegetation Index-SVI, Soil-Adjusted Vegetation Index- SAVI for mangrove map

MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT

VIETNAM FORESTRY UNIVERSITY

STUDENT THESIS

COMPARING VEGETATION INDICES FOR MANGROVE FOREST MAPPING USING REMOTELY SENSED DATA IN KIEN THUY AND DO SON DISTRICT, HAI PHONG CITY

Major: Natural Resources Management (Advanced Curriculum) Code: D850101

Faculty: Forest Resources and Environmental Management

Student: Ha Duc Thien Student ID:1053091718 Class: K55 Natural Resources Management Course: 2010 - 2014

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

Supervisor: Dr Nguyen Hai Hoa

Hanoi, November 2015

ACKNOWLEDGMENTS

“People do not lack strength; they lack will”

Victor Hugo

I would like express my gratitude to my supervisor, Dr Nguyen Hai Hoa for his

support, guidance and encouragement throughout the process of this study

My sincere appreciation to the members of my teammate whose loyalty and creative endeavor made possible this paper

ABSTRACT

The use of vegetation indices of remote sensing data in vegetation mapping has been long recognized However, the accuracy of mapping through the use of vegetation indices model has limitations, and has so far not been investigated This study compared performance of the vegetation indices (Normalized Difference Vegetation Index- NDVI, Simple Vegetation Index-SVI, Soil-Adjusted Vegetation Index- SAVI) for mangrove mapping in Kien Thuy district and Do Son county, Hai Phong city Landsat Image was used as a primary data to derive mangrove vegetation class from three vegetation indices

models A total of three mangrove habitat categories were detected consisting of kandelia Obovate and Sonneratia Caseolaris, Mixed - Kandelia Obovate and Sonneratia Caseolaris, Mixed Sonneratia Caseolaris The accuracy assessment of vegetation indices

were ranged from 68.3% to 75.8% The results indicated that the SAVI was the best index for mangrove mapping compared to other indices with accuracy of 75.8% and able to determine three mangrove classes

ii

KEY WORDS

Vegetation indices, Landsat Image, mangrove mapping performance, accuracy assessment

ACRONYMS

TABLE OF CONTENTS

ABSTRACT i

KEY WORDS ii

ACRONYMS ii

TABLE OF CONTENTS iii

LIST OF TABLES vi

LIST OF FIGURES vii

Chapter I Introduction 1

Chapter II Literature Review 4

2.1 Overview of using vegetation index for mangrove mapping 4

2.2 Key vegetation indexes for coastal mangrove mapping 5

2.3 Significance of study site 14

Chapter III Study goals, Objectives and Methodology 15

3.1 Study goals and Objectives 15

3.1.1 Study goal 15

3.1.2 Study objectives 15

3.2 Study scope 15

3.3 Methodology 16

3.3.1 Investigation and determination of coastal mangrove species composition and its habitat 16

iv

3.3.2.1 Image pre-processing 17

3.3.2.2 Image processing 18

3.3.2.3 Calculating vegetation indices for mangrove classification 19

3.3.2.4 Accuracy assessment 20

3.3.2.5 Post classification 22

3.3.2.6 Mangrove mapping 22

3.3.2.7 Assessing and comparing different kinds of vegetation indices for mangrove mapping 22

3.3 Quantifying spatial dynamics of coastal mangroves in study areas during period 2010 – 2014 22

Chapter IV STUDY SITE, NATURAL AND SOCIOECONOMIC FEATURES 23

4.1 Natural characteristics 23

4.1.1 Geographical location 23

4.1.2 Topography 24

4.1.3 Climate 24

4.1.4 Hydrology 24

4.1.5 Natural resources 25

4.2 Socioeconomic conditions 25

4.2.1 Population 25

4.2.2 Economy 26

4.2.3 Ecological and economic values of mangroves 26

Chapter V RESULTS AND DISCUSSION 29

5.1 Spatial distribution and structures of coastal mangroves in study sites 29

5.2 Comparison of different kinds of vegetation indces for mangrove classification 33

5.3 Dynamics of coastal mangroves during 2010- 2014 37

5.3.1 Thematic maps and dynamics of coastal mangroves 37

5.3.2 Key drivers of coastal mangrove changes from 2010 to 2014 41

Chapter VI GENERAL CONCLUSION, LIMITATION AND FURTHER STUDY 43

6.1 Conclusions 43

6.2 Limitations and further study 43

REFERENCES 45

vi

LIST OF TABLES

Table.3.1 Landsat data used this study 18

Table 5.1: Synthesis of average mangrove structure characeteristics 32

Table.5.2.1 Values of vegetation Indices for mangrove classification 33

Table 5.2.2 Accuracy assessment of image classified using SVI in 2014 34

Table.5.2.3 Accuracy assessment of image classified using NDVI in 2014 35

Table.5.2.4 Accuracy assessment of image classified using SAVI in 2014 36

Table.5.2.5.Summarization of vegetation indices for mangrove classification 37

Table 5.3.1: The extent of coastal mangroves in the study areas (ha) 38

Table 5.3.2: Dynamic of mangroves during period 2010 -2013 39

Table 5.3.3: Dynamic of mangroves during period 2013 -2014 39

LIST OF FIGURES

Fig 3.1: Clipped images of study sites in Hai Phong: (a) image in 2010, (b) image in 2013,

(c) image in 2014 19

Fig.4.1 Study sites in Hai Phong where: (a) Viet Nam map, (b) Hai Phong city map (c) Selected sites as Kien Thuy District and Do Son County 23

Fig 5.1: Species distribution of coastal magroves in Bang La and Dai Hop, Hai Phong 29

Fig.5.2 Mangrove species in different vegetation indices: (a) SVI – Simple Vegetation Index; (b) NDVI- Normalized Difference Vegetation Index; (c) SAVI- Soil Adjusted Vegetation Index 33

Fig 5.3.1: Distribution of mangrove extents during the period 2010 - 2014 38

Fig 5.3.2: Spatial dynamics of coastal mangroves in study sites during two periods 40

Fig 5.3.3: Fluctuation of mangroves area in study area 41

Chapter I Introduction

Mangrove forests appear in the inter-tidal zones along the coast in most tropical and semi-tropical regions (Tuan, Oanh et al., 2002).They are among the most important and productive of ecosystems and provide habitat for wildlife (Wolanski, Brinson et al., 2009) Mangroves play an important role in coastal zones and can reduce damage from the effects of tsunamis The most obvious evidence can be found from the Indian Ocean tsunami of Dec, 2004 (Danielsen, Sorensen et al., 2005) Moreover, mangrove ecosystems stabilize coastlines, clean water, protect land from erosion, and in many cases promote coastal accretion, and provide a natural barrier against storms, cyclones, tidal bores and other potentially damaging natural forces For centuries, mangroves have contributed significantly to the socioeconomic lives of coastal dwellers In addition, they are a source

of timber for fire-wood and provide building materials, charcoal, tannin, food, honey, herbal medicines, and other forest products (Hong and San, 1993) Importantly, mangrove forests are amongst the most carbon-rich ecosystems in the tropics (Donato, Kauffman et al., 2011) and are recognized as performing a vital role in climate change mitigation thanks

to “blue carbon“ storage (Pendleton L, 2012)

Despite their enormous socio-economic value, mangrove ecosystems are under severe threats High population growth, and migration into coastal areas, has led to an increased demand for their products This situation is further exacerbated by insufficient governance, poor planning, and un-coordinated economic development in coastal zones Globally more than 3.6 million hectares of Mangroves have been lost since 1980, and Asia has suffered the greatest loss of 1.9 million hectares (FAO 2007)

Like many other countries in Southeast Asia, the mangrove areas in Vietnam have decreased markedly In Vietnam, it is estimated that the area of mangrove forests was about 400,000 hectares in the early 20th century However, this area has declined dramatically during the past 50 years (Tuan, Munekage et al., 2003) In northern parts of Vietnam, from Mong Cai to Do Son, throughout the periods 1964-1997, mangrove area decreased by 17,094 ha In the Red River plain, the loss of mangrove was 4,640 ha from

1975 to 1991 followed by a decrease of 7,430 ha in 1993 (NEA 2003) Despite government and international efforts in mangrove restoration programs during the 1990„s, mangrove forests on the Northern coast of Vietnam have declined significantly due to aquaculture Nevertheless, mangrove forests in the protected zone are well managed thanks to community-based forest management (Dat and Yoshino et al., 2013)

Hai Phong city with a length of 125 km of sea dykes, is one of the potential area for local mudflats and mangroves From 1989 to 2013, Hai Phong lost 281 hectares of mangrove forests over ten years and gained roughly 355 hectares after twenty four years The annual rate of mangrove loss in Hai Phong was approximately 23 hectares during the first period 1989-2001 From 2001 to 2013, mangrove gained about 53 hectares per year This was due to the Doi Moi economic reform in Vietnam that was established in 1986 The Vietnamese economy was transformed into a market economy Shrimp farming for export was encouraged and promoted by the government As a consequence, mangrove areas were converted to shrimp aquaculture because of the high benefit from shrimp exports (Tuan et al., 2003) On the other hand, from 2001 to 2013, mangrove forests increased slightly since some coastal districts had a good mangrove conservation system in place thanks to community-based forest management in cooperation with local authorities This situation occurred in Hai Phong after three severe tropical cyclones named Washi,

areas there was no damage; however, areas converted to shrimp aquaculture from mangrove forest by local people were devastated in 2005 After 2005, people realized the important role of mangrove in protecting the dyke system and their livelihood They planted mangrove again in vulnerable areas to defend against the typhoons (Dat and Yoshino et al., 2013)

In order to support coastal management and planning programs, mapping mangrove

is a great need for effective measures on inventory of coastal mangrove areas and aquaculture sites and change detection Mangrove maps can be made from investigation in-situ or analyzing from remotely sensing images and GIS techniques

In fact, many studied results indicated that images of Landsat has been for mangrove habitat mapping with different image processing techniques, including Vegetation index (Ramsey and Jensen ,1996 ; Green et al ,1998; Wang et al ,2004 ; Campbel , 1996 ; Baret and Guyot , 1991; Ismail et al., 2010) in which , the index most commonly used is the Normalized Different Vegetation Index (Akamar et al., 2009), other indices based on the same two spectral bands (NIR-near infrared and VIS-visible usually red) have been developed , mostly with the aim of reducing the sensitivity of the index to factors such as soil background , canopy architecture , atmospheric condition However, the accuracy of mangrove mapping through the use of vegetation index models has limitations

In this study, we propose three vegetation indices in comparison consisting of SVI (Simple Vegetation Index), NDVI (Normalized Difference Vegetation Index) and SAVI (Soil-adjusted Vegetation Index), in order to map mangrove forest for the Kien Thuy and

Do Son coasts, Hai Phong, Vietnam using multi-temporal image

Chapter II Literature Review

2.1 Overview of using vegetation index for mangrove mapping

Since the launching of the fist remote sensing satellite in 1970‟s, there have been tremendous efforts made in mapping mangrove areas The advancement of image processing established many techniques and models for mapping vegetation including vegetation indices models approaches (Ismail et al., 2010) Vegetation index (VI) was formed from combination of several spectral values that were mathematically recombined

in such a way as to yield a single value indicating the amount or vigor of vegetation within

a pixel (Campbel, 1996) Vegetation indices (VIs) are more sensitive than individual bands

to vegetation parameters (Baret and Guyot, 1991) and are strongly affected by two main factors:

Effects of soil background on VI:

Until the soil is fully covered by vegetation, the soil background will influence the

VI For incomplete canopies, the wetting of a previously dry soil can cause a change in VI The change is further complicated by the fact that transmission of light through vegetation

is considerably greater in the NIR than in the Red band Both the ratio and the linear combination classes of VI rely on the existence of the soil baseline in Red and NIR wavelength space for soil normalization (Huete, 1998) The intercept of this line is close

to, but does not pass through the origin and there is usually some scatter of soil points away from the principal soil line Such secondary soil influences are most noticeable with Red and yellow colored soils (Kauth and Thomas, 1976).These two factors affect the

discrimination of low amounts of vegetation from bare soil, and are significant in arid regions and in the early stages of vegetation growth (Huete et al., 1984)

Effect of canopy architecture on VI:

The architecture of a vegetation canopy determines the directions that radiation will

be reflected from plant surfaces The vertical elements of an erectophile canopy trap reflected radiation within the canopy, with a corresponding reduction in the amount reflected vertically towards a nadir-oriented radiometer The opposite is true for a planophile canopy The horizontal leaves reflect more in the vertical direction and less is trapped within the canopy (Bunnik, 1978) A nadir-pointing sensor can receive 20-30% more reflected radiation from a planophile canopy than from an erectophile canopy

Canopy architectural effects on VI were examined by Jackson and Pinter (1986) They showed that during the period of peak green vegetation densities, the RVI was about 30% higher for the planophile canopy On the other hand, the PVI was about 30% higher for the planophile canopy than for the erectophile canopy Thus, within a scene that contains similar amounts of vegetation, but different geometries, different VI will respond differently to canopy architecture At first glance, this complicating feature would diminish the value of VI In fact, it adds a new perspective that could possibly be exploited to provide otherwise unobtionable information on canopy architectural feature within a scene

2.2 Key vegetation indexes for coastal mangrove mapping

To understand how vegetation indexes are designed, it is essential to know some concepts rerated to influence of soil and the use of the soil line and vegetation iso-line At this point is useful to introduce the different kind of vegetation indices that have developed over the years Some of the indices have developed considering that all vegetation iso-line converge at a single point These indices are called “ratio – based “and measure the slope

of the line between the point of convergence and the soil line When the vegetation iso-line

is considered parallel to soil line, the indices are called “perpendicular” vegetation indices

Ratio - Based VIs: Ratio - based VI‟s computations are done using data acquired

in visible red and near IR bands These values indicate both the status and abundance of green vegetation cover and biomass This is the first level of classification and aid in delineating areas under vegetation from non-vegetation

Simple Vegetation Index (SVI): RATIO vegetation indices (Rose et al., 1973)

separate green vegetation from soil background by dividing the reflectance values contained in the near IR band (NIR) by those contained in the red band (R)

SVI = NIR / RED

This clearly shows the contrast between the red and infrared bands for vegetated pixels with high index values being produced by combinations of low red (because of absorption by chlorophyll) and high infrared (as a result of leaf structure) reflectance Ratio value less than 1.0 is taken as non-vegetation while ratio value greater than 1.0 is considered as vegetation The major drawback in this method is the division by zero Pixel value of zero in red band will give the infinite ratio value To avoid this situation Normalized Difference Vegetation Index (NDVI) is computed

Normalized Difference Vegetation Index (NDVI): NDVI overcomes the

problem of Ratio method (i.e division by zero) It was introduced in order to produce a spectral VI that separates green vegetation from its background soil brightness (Rose et al., 1974) and is given by,

NDVI = (NIR - RED) / (NIR + RED)

This is the most commonly used VI due to the ability to minimize topographic effects while producing a linear measurement scale ranging from –1 to +1.The negative value represents non vegetated area while positive value represents vegetated area

Ratio Vegetation Index (RVI): The ratio vegetation index is the reverse of the

standard simple ratio (Richerdson et al., 1977),

RVI= RED/NIR

The range for RVI extends from 0 to infinity The ratio value less than 1.0 is taken

as vegetation while value greater than 1.0 is considered as non-vegetation area

Perpendicular VIS: The main objective of the perpendicular vegetation index is

to cancel the effect of soil brightness in cases where vegetation is sparse and pixels contain

a mixture of green vegetation and soil back ground These indices are evaluated on the basis of soil line intercept concept The soil line is a hypothetical line in spectral space that describes the variation in the spectrum of bare soil in the image The soil line represents a description of the typical signatures of soils in red/near-infrared bi-spectral plot It is obtained through liner regression of the infrared band against the red band for sample of bare soil pixels Pixels falling near the soil line are assumed to be soils while those far away are assumed to be vegetation Equation of the soil lines is given below,

Y1 = 0.841333x + 10.781234 (red band independent variable)

Y2 = 0.985684x + 9.501355(infra-red band as independent variable)

Perpendicular Vegetation Index (PVI): PVI uses the perpendicular distance

from each pixel co-ordinate to the soil line and this was derived to define vegetation and non-vegetation for arid and semi-arid region (Richerdson et al., 1977) The pixels, which are close to soil line, are considered as non-vegetation while pixels, which are away from

soil lines, represent vegetation PVI values for data taken at different dates require an atmospheric correction of data, as PVI is quite sensitive to atmospheric variations This can

be defined as:

PVI= sin (a) NIR – cos (a) Red

Where, a = angle between the soil line and the NIR axis

√ Where, (x1, y1) is the co-ordinate of the pixel and (x2, y2) is the coordinate of soil line point that is perpendicular to pixel co-ordinate

Perpendicular distance less than 7.0 is taken as non-vegetation area while greater than 7.0 is taken as vegetation area

Perpendicular Vegetation Index 1 (PVI1): It was noticed that original PVI

equation is computationally intensive and does not discriminate between pixels that fall to the right or left of the soil line (i.e water from vegetation) Given the spectral response pattern of vegetation in which the infrared reflectance is higher than the red reflectance, all vegetation pixels will fall to the right of the soil line In some cases a pixel representing non-vegetation (e.g water) may be equally far from the soil line but it will fall left side of the soil line

In PVI the water pixel will be assigned a high vegetation index value PVI1 assigns negative values to those pixels, which can be delineated from vegetation The mathematical equation for PVI1 (Perry et al., 1984) is written as,

√

Where, NIR: reflectance in the near infrared band, RED: reflectance in the red band , a : intercept of the soil line, b : slope of the soil line

Infrared band is taken as the independent variable and the red band as dependent variable for regression analysis Perpendicular distance less than 6.5 is taken as non-vegetation area while greater than 6.5 is taken as vegetation area

Perpendicular Vegetation Index 2 (PVI2): In PVI2, Red band is taken as

independent variable over infrared dependent variable for regression analysis (Bannari et al.,1996), given importance to the red band with the intercept of soil line Mathematically PVI2 can be represented as,

√

√ Where, a: intercept of the soil line, b: slope of the soil line

Here, pixels having less than –95.0 are grouped as non-vegetation area

Perpendicular Vegetation Index 3 (PVI3): PVI3 is improved version of PVI,

where red band is taken as independent variable on regression analysis and special attention was given to avoid the negative results (Qi et al.,1994) PVI3 can be defined as,

PVI3= a pNIR – b pRED

Where, a: intercept of the soil line, b: slope of the soil line, pNIR: reflectance in the near infrared band, pRED: reflectance in the visible red band

Difference Vegetation Index (DVI): DVI weigh up the near-infrared band by the

slope of the soil line (Richerdson and Wiegand., 1977) and is given as :

DVI= gNIR – RED

Where, g: the slope of the soil line

Similar to the PVI1, with the DVI, a value of zero indicates bare soil, values less

than zero indicate non vegetation and greater than zero indicates vegetation

Weighted Difference Vegetation Index (WDVI): Like PVI, WDVI is very

sensitive to atmospheric variations (Richerdson et al., 1977) and can be presented as,

Where, NIR : reflectance of near infrared band, RED: reflectance of visible red

band, g: slope of the soil line

Although simple, WDVI is as efficient as most of the slope based VI‟s The effect

of weighting the red band with the slope of the soil line is the maximization of the

vegetation signal in the near-infrared band and the minimization of the effect of soil

brightness

Soil noise

Soil reflectance spectra depend on type of soil The vegetation indices computed

earlier assume that there is a soil line, where there is a single slope in red-NIR space

However, it is often the case that there are soils with different red-NIR slopes in a single

image Also, if the assumption about the iso-vegetation line (parallel or intercepting at the

origin) is not exactly right, changes in soil moisture (which move along iso-vegetation

lines) will give incorrect answers for the vegetation index The problem of soil noise is

most acute when vegetation cover is low The following groups of indices like SAVI,

TSAVI1, TSAVI2, MSAVI1, MSAVI2 attempt to reduce soil noise by altering the

behavior of the iso-vegetation lines All of them are ratio-based, and the way that they

These indices reduce soil noise at the cost of decreasing the dynamic range of the index These indices are slightly less sensitive to changes in vegetation cover than NDVI (but more sensitive than PVI) at low levels of vegetation cover These indices are also more sensitive to atmospheric variations than NDVI (but less so than PVI)

Soil-adjusted Vegetation Index (SAVI): SAVI is intended to minimize the

effects of soil background on the vegetation signal by incorporating a constant soil adjustment factor L in the denominator of the NDVI equation (Huete, 1988) L varies with the reflectance characteristics of soil (i.e color and brightness) The L factor chosen depends on the density of the vegetation For very low vegetation L factor can be taken as 1.0 while for intermediate it can be taken as 0.5 and for high density 0.25 The best L value

to select is where the difference between SAVI values for dark and light soil is minimal For L=0, SAVI equals NDVI For L=100, SAVI approximates PVI Mathematically SAVI

is defined as,

SAVI = {(NIR - RED) / (NIR+RED+ L)} * (1+L)

Where, NIR: near-infrared band, RED: visible red band, L: soil adjustment factor

Multiplicative term (1+L) present in SAVI (and MSAVI) is responsible for vegetation indices to vary from –1 to +1 This is done so that both vegetation indices reduce to NDVI when the adjustment factor L goes to zero

Transformed Soil-adjusted Vegetation Index (TSAVI 1): SAVI concept is exact

only if the constants of the soil line are a = 1 and b = 0, where a is slope of the soil line and

b is y-intercept of the soil line As it is not generally the case some modification was needed in SAVI By taking into the consideration of PVI concept (Baret et al., 1989), SAVI is modified as TSAVI1 This index assumes that the soil line has arbitrary slope and intercept, and it makes use of these values to adjust the vegetation index and is written as:

Where, NIR: reflectance in near infrared band, RED: reflectance in red band, a: slope of the soil line, b: intercept of the soil line, X: adjustment factor which is set to minimize soil noise

Red band is taken as independent variable for regression analysis Ratio value less

as non-vegetation while greater than –9.0 is taken than –9.0 is taken as vegetation With some resistance to high soil moisture, TSAVI1 could be very good candidate for use in semi-arid regions TSAVI1 was specifically designed for semi-arid region and does not work well in areas with heavy vegetation

Transformed Soil-adjusted Vegetation Index (TSAVI2): TSAVI2 is modified

version of TSAVI1 which was readjusted with an additive correction factor of 0.08 to minimize the effects of the background soil brightness (Baret et al., 1989), and is given by,

Red band is taken as independent variable for regression analysis and is given preference with soil line intercept

Modified Soil-Adjusted Vegetation Index 1 (MSAVI1): The adjustment factor L

for SAVI depends on the level of vegetation cover being observed which leads to the circular problem of needing to know the vegetation cover before calculating the vegetation index which is what gives you the vegetation cover MSAVI is the Modified Soil Adjusted Vegetation Index (Qi et al.,), and provide a variable correction factor L The correction

Where, L: 1 - 2 * s * NDVI * WDVI, s: slope of the soil line, NDVI: Normalized Difference Vegetation Index, WDVI: Weighted Difference Vegetation Index, 2: Used to increase the L dynamic range, range of L = 0 to 1

Modified Soil-Adjusted Vegetation Index 2 (MSAVI2): MSVI2 was derived

based on a modification of the L factor of the SAVI (Qi et al.,) SAVI and MSVI2 are intended to correct the soil background brightness in different vegetation cover conditions Basically, this is an iterative process and substitute 1-MSAVI (n-1) as the L factor in MSAVI (n) and then inductively solve the iteration where MSAVI (n) = MSAVI (n-1) MSVI2 uses an inductive L factor to:

•Remove the soil “noise” that was not canceled out by the product of NDVI by WDVI; and

•Correct values greater than 1 that MSAVI1 may have due to the low negative value of NDVI * WDVI Thus its use is limited for high vegetation density areas

The general expression of MSAVI2 is,

2.3 Significance of study site

Hai Phong is a coastal city with a total of 125km of seacoast Coastal mangroves not only reduce consequences from natural disaster but also bring huge economic benefit for local people Using Remote Sensing & GIS (Geographic Information System) as a tool would enhance effectiveness of coastal mangroves management Up to date, there is no published document on applying vegetation index for coastal mangrove in Hai Phong city,

in particular in Kien Thuy and Do Son Thus, this study would be significantly contribution

to technical enhancement for mangrove exploring and mapping in Hai Phong

Chapter III Study goals, Objectives and Methodology

3.1 Study goals and Objectives

In order to reach the main goal of study, three specific objectives are given as below:

Objective 1:To investigate the spatial distribution and structures of coastal

mangrove species in Dai Hop and Bang La communes, Hai Phong

Objective 2: To determine different vegetation indices for classifying mangroves

in Dai Hop and Bang La communes, Hai Phong

Objective 3: To quantify the spatial dynamics of coastal mangroves in study areas

3.3 Methodology

3.3.1 Investigation and determination of coastal mangrove species composition and its habitat

Field survey: The field survey was divided into two phases Calibration data were

collected in May 2014, accuracy data in September 2014 A total of 335 sample sites in square shape (30m * 30m - equal pixel size) were established Species composition, tree height , tree density and crown diameter were recorded at all sites

a tape measure

Tree height and crown diameter were measured using a 5 m telescopic pole

Tree density was measured by counting the number of tree trunks at breast height

When a tree forked beneath breast height (1.3 m) each branch was recorded as

a separate stem The location of each field site was determined using GPS with a probable circle error of 2-5 m

Canopy closure/coverage (%): An estimation of canopy closure is made

using the densitometer approach This device is made from piece of 2.5 cm diameter duct pipe approximately 40 cm long with cross-hairs added at both ends using fine wire threaded evenly across the diameter of the tube (Duke et al., 2010; Hai-Hoa et al., 2013) The densitometer is held vertically and an estimate made of the percentage of the view through the duct pipe covered by sky, or less or more than 50% of a leaf, branch and tree trunk (recorded as 0, 0.5 or 1 respectively) Readings are taken every meter along the transect length within a 100 m2 plot, resulting in about 100 readings

Determination of Mangrove Habitat Categories: A habitat classification was

developed for the mangrove areas of the Dai Hop and Bang La using hierarchical agglomerative clustering with group-average sorting applied to the calibration data The calibration data were transformed in order to weight the contribution of tree height , crown diameter and density more evenly with species composition (the range of data was an order of magnitude higher for density, height, crown diameter and would cluster accordingly) 85% level of similarity was used for cluster analysis Categories were described in terms of mean (or median) species composition (percent species), mean (or median) tree height , mean (or median) tree density and mean (or median) crown diameter

These habitat categories would use to direct the image classification of the Landsat data, and the collection of accuracy data in September 2014

3.3.2 Calculation of different types of vegetation indices used to classify mangroves

Vegetation indices are calculated using near Red and Infrared bands of Landsat Image Mangrove classes were identified and edited with reference to field data, then merged into a single mangrove category Assessment accuracy and post-processing are the last procedures before exporting map

GIS applications to perform VI‟s classification and mapping

3.3.2.1 Image pre-processing

Remote sensing data for study

Landsat satellite images are used to calculate vegetation indices and to classify coastal mangroves (Table 3.1)

Table.3.1 Landsat data used this study

2010 LE71260462010361EDC00 27/12/2010 30(15) m 126/46

2013 LC81260452013281LGN00 08/10/2013 30(15) m 126/45

2014 LC81260462014284LGN00 11/10/2014 30(15) m 126/46

Source: http://earthexplorer.usgs.gov

Convert DN values to TOA reflectance:

At this stage, atmospheric and radiance corrections were applied to the image for

improving the quality of vegetation classification by using the Spatial Analyst Tools in

ArcGIS 10.1 as below:

ArcToolbox => Spatial Analyst Tools => Map Algebra => Raster Calculator:

Lλ =MLQcal +AL

Wherre ML: Band-specific multiplicative rescaling factor from the metadata

(radiance_Mult_Band_x, where x is the band number)

AL: Band-specific additive rescaling factor from the metadata (Radiance_add_band_x,

where x is the band number)

QCal: Quantized and calibrated standard product pixel values (DN)

3.3.2.2 Image processing

Composite bands: The image was corrected to UTM zone 48 N, WGS 84

band and Green band of the Landsat Contrast, brightness and anomalies are verified before using as original image data

Clipping the image: Because the study area forms only a part of the image, it is

necessary to cut separating the study area A file containing the study area boundaries are used to cutting areas of research topics from newspaper photographs

Fig 3.1: Clipped images of study sites in Hai Phong: (a) image in 2010, (b) image

in 2013, (c) image in 2014

3.3.2.3 Calculating vegetation indices for mangrove classification

To calculate vegetation index, the study uses ArcGIS 10.1 with the extension of Spatial Analysis Tools

Step 1: Calculation vegetation index uses Red band (0.636 – 0.673 µm) and NIR

band (0.851 – 0.879 µm) of the image

Calculation of NDVI: NDVI = (NIR - RED) / (NIR + RED)

Calculation of SVI: SVI = NIR / RED