Jiraporn_Kongwongjanx - APAN

molassesitalianAI and Robotics

Nov 6, 2013 (3 years and 9 months ago)

75 views

Comparison of
vegetation indices for
mangrove mapping
using THEOS data

Jiraporn Kongwongjun,

Chanida Suwanprasit

and Pun Thongchumnum

Faculty of Technology and Environment,
Prince of Songkla University,


Phuket Campus

APAN
-
33rd Meeting

1

Outline

1.
Introduction

2.
Objectives

3.
Study area

4.
Methodology

5.
Result

6.
Conclusion

7.
Acknowledgement

2

The importance of mangroves

Mangrove forests are
useful
as fishing areas, wildlife reserves, for recreation,
human habitation, aquaculture and
natural
ecosystem.

1

4

2

6

3

5

3

Mangrove vegetations

1

4

2

6

3

5

(a)
Rhizophora mucronata

Poir I

(b)
Rhizophora apiculata

Blume

(c)
Sonneratia ovata

Backer

(d)
Rhizophora Ceriops
Decandra

(e)
Rhizophora Bruguiera
s.

(Department of marine and coastal
resource,
2
011
)

4

Vegetation indices

1

4

2

6

3

5



The remote sensing is applicable for


mangrove mapping.



The
vegetation indices
(
VIs) in


forest
areas have been widely used


and provide accurate classification
.



Different VIs is suitable for different


vegetation cover.


5

Objectives

1

4

2

6

3

5



To classify mangrove
and
non
-
mangrove



areas.




To
find out a suitable vegetation index for


identifying mangrove
area.



6

Study area

1

4

2

6

3

5

Pa Khlok sub
-
district,


Phuket, Thailand

7

Study area

1

4

2

6

3

5

source: www.technicchan.ac.th
,
2011

source: http://cccmkc.edu.hk

8

Methodology

1

4

2

6

3

5

Input THEOS data


Pre
-
Image Processing

Image classification

Post classification

Compare Image

Output mapping data

ROI

Training

Test

5 VIs



NDVI


SR


SAVI


PVI


TVI



unsupervised

supervised

K
-
mean


Visual Interpretation

9

THEOS Satellite

1

4

2

6

3

5

Description

MS

Spectral bands and resolution

4 multispectral (15 meters)

Spectral ranges

B1 (blue) : 0.45
-
0.52 µm

B2 (green) : 0.53


0.60 µm

B3 (red) : 0.62


0.69 µm

B4 (NIR) : 0.77


0.90 µm

Imaging swath

90 km.

Image dynamics

8 bits
-
12 bits

Absolute localization accuracy (level
1B)

< 300 m (1 s)

Off
-
nadir viewing

±
50
°

(roll and pitch)

Signal to Noise Ratio

>100

(Pitan, 2008)

10

1

4

2

6

3

5

Band1: Blue

0.45
-
0.52
µm

Band2: Green

0.53


0.60
µm

Band3: Red

0.62


0.69
µm

THEOS Spectral bands

Band4: NIR

0.77


0.90
µm

11

1

4

2

6

3

5

Selection of ROIs

ROIs

Training

pixels

(50%)

Test

pixels

(50%)

Mangrove

691

691

Non
-
mangrove



water



cloud on water



cloud on land



f
潲敳t



慧a楣i汴畲l


佴桥牳


1,364

13

132

1,118

387

88


1,364

13

132

1,118

387

88

Total

3,661

3,661

12

1

4

2

6

3

5

ROIs Table

Class

Mangrove

Cloud


(water)

Cloud

(land

Forest

Agriculture

water

Others

Mangrove

-

2.00

1.98

1.59

1.33

2.00

1.99

Cloud

water

2.00

-

2.00

2.00

2.00

1.99

1.99

Cloud

land

1.98

2.00

-

1.98

1.96

2.00

1.98

Forest

1.61

2.00

1.99

-

1.72

1.99

1.99

Agriculture

1.29

2.00

1.97

1.71

-

1.99

1.99

water

2.00

1.99

2.00

1.99

1.99

-

1.99

Others

1.99

1.99

1.98

1.99

1.99

1.99

-

Training Sample ROI

Test Sample ROI

13

1

4

2

6

3

5

Class

Formulas

Authors

Normalized Different Vegetation Index
(NDVI)

(Pearson and Miller, 1972)

Simple Ratio (SR)

(Pearson and Miller, 1972)

Soil Adjusted Vegetation Index (SAVI)

(
Huete
, 1998)

Perpendicular Vegetation Index (PVI)

(Richardson and

Wiegand,1977)

Triangular Vegetation Index (TVI)


0.5(120(NIR
-
G) )
-
200(R
-
G)

(
Broge

& Leblanc, 2000)

RED
NIR
R)

(NIR
R)

-

(NIR

L)


R


(NIR
L)
(1

R)

-

(NIR



2
,
,
2
,
,
)
(
)
(
NIR
V
NIR
S
R
V
R
S







5 Vegetation Indices

14

1

4

2

6

3

5

Vegetation Indices

NDVI

SR

SAVI

PVI

TVI

15

1

4

2

6

3

5

Image Classification

K
-
mean

MLC+NDVI

MLC+SR

MLC+SAVI

MLC+TVI

MLC

MLC+PVI

Classification 2
classes
:
mangrove
and
non


mangrove areas

Unsupervised

Supervised

Yellow
= Non
-
mangrove

Blue
=

Mangrove

16

1

4

2

6

3

5

Overall accuracy

Classified

Overall accuracy

Kappa coefficient

Maximum Likelihood (MLC)

96.46%

0.9522

MLC+ NDVI

96.78%

0.9565

MLC+ SR

96.78%

0.9565

MLC + SAVI

96.78%

0.9565

MLC + PVI

95.67%

0.9417

MLC + TVI

95.30%

0.9364

17

1

4

2

6

3

5

Conclusion

18



NDVI, SR and SAVI are the
best indices
between


mangrove and non
-
mangrove forests
with
96.78%
overall


accuracy.



THEOS with 15 m resolution is appropriate for visual


interpretation. However, spectral resolution of 4 bands


seems to give limited vegetation classification.





1

4

2

6

3

5

Acknowledgement



Faculty of Technology and Environment, Prince of
Songkla



university,
Phuket

campus, providing
invaluable assistance


during work




Geo
-
Informatics and Space Technology Development


Agency organization (
GISTDA
)




UniNet




Adviser and co
-
adviser in particular to
Dr.Chanida



Suwanprasit

and
Dr.Pun

Thongchumnum

who
give suggestion


and
Dr.Naiyana

S
richai

and my graduate
friends for


encouragement.


19

1

4

2

6

3

5

References


Pitan

Singhasneh

(
2
011
). "

THEOS Satellite Data Service
"
<
http://www.gisdevelopment.net/technology/rs/mwf09_theos.htm> ( 10 February

2012
)


Cccmkc

University

(
2
011
). "
Mangrove in
Phuket
, Thailand
"
<
http://cccmkc.edu.hk/~kei
-
kph/Mangrove/mangrove_page%201.htm> ( 10 February

2012
)


Huete

A. (1988). “A soil
-
adjusted vegetation index (SAVI).”
Remote Sensing of Environment
, 25
(3), 295
-
309.


Richardson A. J. and
Wiegand

C. L. (1977). “
Distinguishing vegetation from soil background
information(by gray mapping of
Landsat

MSS data” Photogrammetric Engineering and Remote
Sensing
., 43(12), 1541
-
1552.


Pearson, R. L. and Miller, L. D. (1972). “Remote mapping of standing crop biomass for estimation
of the productivity of the
shortgrass

prairie, Pawnee National Grasslands, Colorado” Proceedings
of the 8th International Symposium on Remote Sensing of the Environment II., 1355
-
1379.


Broge
, N. H., & Leblanc, E. (2000). “
Comparing prediction power and stability of


broadband and
hyperspectral

vegetation indices for estimation of green leaf area


index and canopy chlorophyll density
”. Remote Sensing of Environment, 76,


156−172.


Department of marine and coastal resource
. (
2
011
). "

Research Paper 14th Mangrove National
Seminar
"
<

http://issuu.com/mffthailand/docs/mangrove14th > ( 10 February

2012
)

20

THANK YOU

FOR

YOUR ATTENTION

Jiraporn Kongwongjun,

Chanida Suwanprasit

and Pun Thongchumnum

Faculty of Technology and Environment,
Prince of Songkla University,


Phuket Campus

APAN
-
33rd Meeting

21