Mapping seagrass and seaweed Mapping seagrass and seaweed

Mapping seagrass and seaweed Mapping
seagrass and seaweed
beds in NOWPAP
Teruhisa Komatsu
Atmosphere and Ocean Research Institute, The University of Tokyo
Contents
What is “coastal habitat”?
Important ecological roles of coastal habitats
Mapping methods of coastal habitats
Direct and indirect methods
Mapping by remote sensing Introduction of seagrass mapping
What is habitat?
What is habitat?
Habitat is similar to biotope or biome:
An area that is uniform in environmental
An area that is uniform in environmental conditions and in its distribution of animals and plants. plants
On land, forests, meadows and marshes are habitats They are called ecological engineers. habitats. They are called ecological engineers
What habitats exist in the sea?
Seaweed and seagrass beds
Seaweed and seagrass beds
Z t
Zostera
marina
i
Sargassum horneri
Zostera caulescens
Ecosystem services of habitats
y
•
•
•
•
•
•
•
•
•
•
•
Gas regulation
Climate regulation
Disturbance regulation
Erosion control and sediment retention
Nutrient cycling
W t t t
Waste treatment
t
Refugia (spawning and nursery grounds )
Food production
Food production
Genetic resources
Recreation
Cultural
Costanza et al. (1997)
Ecological services of coastal biomes in the world
Ecological services of coastal biomes in the world
Biome
Area
(ha)
Total value per ha
Total
value per ha Total global flow Total global flow
value($yr‐1 x109)
($ha‐1 yr‐1)
Estuaries
Estuaries 180
180 22,832
22,832 4,110
4,110 Seagrass/seaweed beds 200 19,004 3,801 62 6,075 375 2,660 1,610 4,283 165 9,990 1,648 Total coastal biome 3,267 4,352 14,217 T i lf
Tropical forest t
1 900
1,900 2 007
2,007 3 813
3,813 Temperate/boreal forests 2,955 302 894 Total forest biome
forest biome
4 855
4,855 970
970 4 707
4,707 Coral reefs Shelf Tidal marsh/mangroves
Constanza et al. (1997)
Coastal habitats at risk
Agricultural pollution
Reclamation
l
i in
i Seto
S
Inland Sea around 1960s
Industrial pollution
Seagrass beds in Seto
Inland Sea
Amelioration of water quality in Seto Inland Sea
Komatsu (1997) Long‐term changes in the Zostera bed area in the Seto Inland Sea (Japan), especially along the coast of the Okayama Prefecture, Oceanologica. Acta, 20, 209‐216.
Law on Environmental Conservation of the Seto Inland Sea
Cumulative reclamation and Zostera marina
Cumulative reclamation and Zostera
Komatsu (1997) Long‐term changes in the Zostera bed area in the Seto Inland Sea (Japan), especially along the coast of the Okayama Prefecture, Oceanologica. Acta, 20, 209‐216.
Counter measure against trawls in seagrass meadows in Okayama Prefecture, Japan
Deployment of artificial reefs and columns obstructing trawl operation Deployment
of artificial reefs and columns obstructing trawl operation
inside the seagrass beds of Ajino Bay in 1976
Deployment
Komatsu (1997) Long‐term changes in the Zostera bed area in the Seto Inland Sea (Japan), especially along the coast of the Okayama Prefecture, Oceanologica. Acta, 20, 209‐216.
Management of coastal habitats
• Monitoring of habitat distributions
Monitoring of habitat distributions
• Protection of habitat areas
• Restoration of habitats
Mapping coastal habitats is indispensable
for sustainable use of coastal resources
Mapping coastal habitats by direct and indirect surveys
direct and indirect surveys
Direct methods
•
•
•
•
Walkingg
Diving
Observation from the ship
p
Grabbing bottom sediments
Direct methods (ground survey)
Characteristics
density estimation,
estimation species identification
assured method
Problems
low efficiency
influence of turbidity of water and
high waves on field survey
Mapping of habitats by indirect methods
th d
• Acoustics
• Optics (satellite remote sensing)
Optical methods
Optical methods
Introduction of seagrass mapping
Lyzenga’s Model
Sun
Esun
Lsi
Li
Satellite
Sea surface
Deep D
water
Bottom surface
Ri
Ki Z
Li = Lsi + Ai・
Ai・Ri
Ri・
・exp
exp((‐Ki
Ki・
・F・Z) (W/m2/sr
sr))
(Lyzenga 1978)
Lyzenga (1978, 1981)
(1978 1981)
Depth‐invariant Index (DII)
• DII = ln( Li‐Lsi
(
))‐[( Ki/Kj
[( / j ))ln( Lj‐Lsj
( j j )]
)] DII = ln( Ai Ri
l ( i i )‐( Ki/Kj
) ( i/ j )ln( Aj
)l ( j Rjj ) )
assuming
g that Ki/Kj=1
j
DII = c + ln (Ri/Rj)
Lyzenga’s model:
Li = Lsi + Ai・
Ai・Ri
Ri・
・exp(
exp(‐
‐Ki
Ki・
・f・Z)
Sagawa et al. (2010) g
(
) International Journal of Remote Sensing, 31, 3051‐3064
Reflectance Index (RI)
•
RI = ( Li ‐Lsi )/( exp( ‐Ki・f・Z ) )
RI = Ai・Ri
Lyzenga’s model: Li
Lyzenga’s model: Li = Lsi + Ai・
= Lsi + Ai・Ri
Ri・
・exp(‐Ki
exp(‐Ki・
・f・Z)
Example of attenuation coeeficient
F=2.18
K= 0.093 m‐1
J l W
Jerlov Water Type Ⅱ‐Ⅲ
T
ⅡⅢ
R l ti b t
Relation between depth
d th and radiance (green band)
d di
(
b d)
Lyzenga’s model: Li Li == Lsi + Ai・Ri・exp
exp(‐
(‐Ki
Ki・F・Z) (W/m2/sr
) (W/m2/sr))
DI Index
誤差行列(Error matrix)
全体の精度(Overall accuracy) = 54 %
Sagawa et al. (2010) International Journal of Remote Sensing, 31, 3051‐3064
BR Index
Sagawa et al. (2010) International Journal of Remote Sensing, 31, 3051‐3064
Mahares
¾
Overall
accuracy
DI)(%)
(DI)
(%)
Overall
accuracy
(BR)(%)
(BR
)(%)
JWT
54.0
90.0
Ⅱ-Ⅲ
Classification with BR Idex is more correct than DI Index
(p < 0.05)
Sensors
A i l photography
Aerial
h
h
Launch of Landsat satellite (1972)
Sensors with spatial resolution (30~100 m) were most widely used
(e.g. Landsat Multispectral Scanner, Landsat TM)
long-time series/inexpensive data
Sensors with better spatial
p
resolution <10 m
(e.g. IKONOS, SPOT, Worldview 2, Geo Eye etc)
better spectral & spatial versatility/extremely expensive data
Launch of ALOS (2006) and stop in April 2011
AVNIR-2
AVNIR
2
resolution: spatial (10 m)/multispectral (420~890nm)/radiometric (8 bits)
Characteristics of Landsat TM
Characteristics of Landsat TM
Band no. Spectral range (µm)
Spatial resolution (m)
1
0.45 - 0.52
30
2
0.52 - 0.60
30
3
0.63 - 0.69
30
4
0.76 - 0.90
30
5 1,
1.55 -41.75
30 to those
Bands
1 2,
2 3 and
are nearly equivalent
6
10.4 - 12.5
120
of ALOS.
2 08
2.08
- 2.35
2 35
30
We 7can analyze
time
series data consisting
of
LANDSAT TM and ALOS from 1972 to 2011.
Characteristics of AVNIR 2 sensor
Characteristics of AVNIR‐2 sensor Swath Width
Spatial Resolution
Wavelength
Quantization
70 km ( at nadir)
10 m (at nadir)
Band 1: 0.42 – 0.50 µm (visible blue)
Band 2: 0.52 – 0.60 µm (visible green)
Band 3: 0.61 – 0.69 µm (visible red)
Band 4: 00.76
76 – 0.89
0 89 µm
m (near infrared)
8 bits
We can compare habitat distributions between images of ALOS AVNIR‐2 and LANDSAT
g
Study Area
Sibu Island, Johor
Malaysia
Raw ALOS AVNIR‐2 Image (29 J l 2008 11:54 am)
(29 July 2008 ‐
11 54
)
Change off corall reeff tto Sargassum
Ch
S
beds from 2005 to 2008
Comparison
C
i
result
lt off ALOS (2008) and
d
LANDSAT ((2005)) over Sibu Island
Detection of succession from live coral
to Sargassum forest
Courtesy of Prof. Ibrahim Seeni at UTM, Malaysia
28
Tentative proposal to NOWPAP
Tentative proposal to NOWPAP
• TTemporal changes in spatial distributions of l h
i
ti l di t ib ti
f
seaweed and seagrass beds with relation to environmental changes such as eutrophication or
environmental changes such as eutrophication or reclamation
• Focusing on changes in not only distributions of seagrass and seaweed beds but also land use
seagrass and seaweed beds but also land use • Selection
Selection of test sites, where they are distributed of test sites where they are distributed
broadly, with time series data of environmental p
parameters
Thank you very much for your attention!