Using SPOT-LANDSAT images for mapping, inventory and monitoring of reefs - Serge Andréfouët - Remote Sensing/ Biological Oceanography University of South.

Using SPOT-LANDSAT images for mapping, inventory and monitoring

of reefs

- Serge Andréfouët -

Remote Sensing/ Biological Oceanography University of South Florida, St Petersburg, USA

Laboratoire de Géosciences Marines et TélédétectionUniversité Française du Pacifique, Tahiti

Lagoon

Anaa Atoll (French Polynesia)SPOT XS-3,2,1

OceanRim

5 km

Marquesas Key (Florida, USA)

LANDSAT 5 TM-3,2,1

5 km

20/30 meters

20/3

0 m

eter

s

20/30 meters

20/3

0 m

eter

s

40/60 meters

40/6

0 m

eter

s

300 meters

300

met

ers

1500 meters

1500

met

ers

Lagoon

Ocean

AtollLagoon

XS3

XS1

XS2

XS3

XS2

XS1

SPOT IMAGES

XS3

XS2

XS1

Pixel P

P

Spectral space

Remotely sensed information

• LwXSi= LwXSb + LwXSw (+ Lwa)

SPOT: XS1, XS2

LANDSAT: TM1, TM2, TM3

• LwXSb related to the “bottom” features

• LwXSw related to the water column features

Spectral discrimination of organismsLwb

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

400 450 500 550 600 650 700Longueur d'onde (nm)

Ref

lect

ance

Living Porites sp.

Coral rubble with turf

Coral rubble with endoliths

Living Montipora sp.

Algal turf

Beer Can

Spectral discrimination

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

400 450 500 550 600 650 700 (nm)

Ref

lect

ance

Sensitivity TM1 Sensitivity XS2Sensitivity XS1

Spectral discrimination

0

0.2

0.4

0.6

0.8

1

1.2

400 450 500 550 600 650 700 (nm)

Ref

lect

ance

SPOT XS1 sensitivity SPOT XS2 sensitivity

0 meters depth

0

0.2

0.4

0.6

0.8

1

1.2

400 450 500 550 600 650 700 (nm)

Ref

lect

ance

SPOT XS1 sensitivity SPOT XS2 sensitivity

Under-water Spectral discrimination

5 meters depth

0

0.2

0.4

0.6

0.8

1

1.2

400 450 500 550 600 650 700 (nm)

Ref

lect

ance

SPOT XS1 sensitivity SPOT XS2 sensitivity

Under-water Spectral discrimination

20 meters depth

Minimum Discernable Unit (MDU)

Size_MDU = PixelSize.(1+2.ErrorLocation)

if ErrorLocation= 1 pixel (pretty good!!)

SPOT MDU= 60 m x 60 m

LANDSAT MDU = 90 m x 90 m

Minimum Discernable Unit (MDU)

CASI image:

PixelSize= 1 meter

2 x 2 m : not enough

4 x 4 m : ok for training

Minimum Discernable Unit (MDU)

MDU= 3 x 3 m MDU= 60 x 60 m

Remotely sensed information

• Lwi= Lwb + Lww (+ Lwa)

• 2 or 3 known measurements: XS1 and XS2

TM1, TM2 and TM3

• 2 unknown variables Lwb and Lww

Haraiki atoll (French Polynesia)

Bathymetric modeling(Lww)

Computed depth Real depth

-30

-25

-20

-15

-10

-5

0

1

Depth

8km

Image of the bottom Scale

“Radiance” scale

“Bottom” reconnaissance(Lwb)

Architecture (forms and dimensions)

Source: Veron (1986)

Massive

Columnar

Free-living

Foliaceous

Encrusting

Branching

Laminar

Hierarchical clustering of the stations

Similarity

field stations

Pure SandSand/Rubble with Isolated-Patches

Reef

Soft Bottom Hard bottom

Dead Living

Pure Rubble

Living coral

What type of habitat can you map with SPOTwith a good accuracy (70%) ?

Depth < 7-8 metersDefinition: coarse

Minimum Discernable Unit= 60 meters x metersBoundary analyses

Transition

Abruptboundaries

Gradient

Patches

Fragmented

Spatial structure of a reef system

• A reef is a complex object, but any part of the reef has a membership degrees in each of the classes

• This membership belongs to [0...1]

• Mapping of membership degrees: fuzzy classification

Is this membership degree useful?

• Mapping

• Habitats boundary analyses

• Acanthaster planci outbreaks

Tiahura

Ocean

Land

Land

Land

Land

Motu

Motu

Motu

Ocean

Ocean

Ocean

Fuzzyclassification

One map for each class of bottom.Mapping of the degree of membership.

Coral

Heterogeneous

Dead structures

1

0

Membership degree:

Tiahura

Ocean

Land

Land

Land

Land

Motu

Motu

Motu

Ocean

Ocean

Ocean

Fuzzyclassification

One map for each class of bottom.Mapping of the degree of membership.

Coral

Heterogeneous

Dead structures

1

0

Membership degree:

Tiahura

Ocean

Land

Land

Land

Land

Motu

Motu

Motu

Ocean

Ocean

Ocean

Fuzzyclassification

One map for each class of bottom.Mapping of the degree of membership.

Coral

Heterogeneous

Dead structures

1

0

Membership degree:

Is this membership degree useful?

• Mapping

• Habitats boundary analyses

• Acanthaster planci outbreaks

Tiahura

2.5 km

Coral

Isolated Patches

Sand

Transitions between bottom types

Lan

d

1

0

Possibility measurement

Lan

d

Lan

d

Lan

d

Is this membership degree useful?

• Mapping

• Habitats boundary analyses

• Monitoring and sampling designs (Acanthaster planci outbreaks)

Location of A. planci infestations in the 80’s (Faure, 1989)

Land

Ocean

0

0.2

0.4

0.6

0.8

1

1.2

400 450 500 550 600 650 700 (nm)

Ref

lect

ance

SPOT XS1 sensitivity SPOT XS2 sensitivity

0 meters depth

What about change detection ?

What about change detection ?

TEKOKOTA

050

100150

200250

300350

400

1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 171

Radiance SPOT

Soft Bottom

Hard Bottom Mixte Bottom

MAROKAU

0

200

400

600

800

1000

1200

1400

1600

1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161 171 181 191 201 211 221

Radiance SPOT

Soft bottom

Mixte bottom

Hard bottom

Histograms of bottom-types in XS1 after bathymetric corrections for 2 atolls

What about change detection ?

Problems in calibration and correction of the images: not enough accurate

Benthos:

• Shifts in living communities : ??????• Change in sediment cover (hurricanes) : ok

Work in the field

Moorea: 20 transects (60m x~1km) for training and control, 6 days, 2 investigators (Yannick Chancerelle, CRIOBE, Moorea),Semi-quantitative (5%, 15%, 25%, >50%) rapid assessment for 4 variables

Atolls: 20 transects, 2 days, 2 investigators

Caveat: Only assessment of the coarse level of habitat without

hierarchical sampling (if not, time x 10) !!!

Work in the image processing lab

Bathymetric correctionFuzzy classification to output membership degreesMapping of the membership degrees

3days - 1week

Conditions:

- user-friendly software does exist

- good control of the software

- good quality of the data (image and field data)

- skilled analyst (if not, time x 10)

Water parameters

Few direct observations. Potentially interesting for atoll lagoons

(phytoplanctonic biomass or suspended matter)

Many indirect observations (the water body is not the target)

rivers run-off, pollution, boundary characterization and residence time

Spatial structure of a reef system and fluxes

Reka-Reka Tepoto Sud Tekokota

Boundary conditions controls: Nutrients limitations Residence time of lagoon waters Recruitment Community structure

Atoll rims typology

aperture 33 %

Wave Exposure Hydrodynamic aperture

South

aperture > 70 %

Structure

y = 2.4608x + 1.4522

R2 = 0.7304

0

0.5

1

1.5

2

2.5

3

3.5

4

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

H_Topex (m)

Flows (m2/s)

Empirical relationships between flows of oceanic waterand wave height for each type of rim

0

50

100

150

200

250

300

350

400

1.5 2 2.5 3 3.5 4

ARUTUAResidence time (days)

H_Topex (m)

0

100

200

300

400

500

600

1.5 2 2.5 3 3.5 4

Marokau

Residence time in atoll lagoons

Distribution of coral reefsEstimated 255.000 km2

(Spalding & Grenfell, 1997)

Global scale

Similarity

field stations

Pure SandSand/Rubble with Isolated-Patches

Reef

Soft Bottom Hard bottom

Dead Living

Pure Rubble

Living coral

Global Coverage: NASA plans to collect ~200 LANDSAT 7 images per day

worldwide:Long-Term Acquisition Plan

(LTAP)

Present coverage of reefs by LTAP

Expected: one cloud-free image per year

Present coverage of prioritized reefs by LTAP(research activities)

Expected: 4-6 cloud-free images per year

Global coverage

Estimation of global distribution of reefs, • without ground-truth, • 2 classes (soft and hard-bottom), • 80% accuracy

A basis for extension of monitoring worldwide, • 6 classes (gradient of soft and hard-bottom), • with ground-truth, • 70% accuracy• Interface with monitoring organizations is required to get training data for image processing

Conclusions

Using SPOT-LANDSAT images for mapping, inventory and monitoring of reefs?

Pragmatic point of view

Mapping: Yes: - Coarse habitats with ground truth

Bathymetric and atmospheric corrections required - Soft/hard bottoms without ground-truth and corrections

- Boundary analyses

Inventory: Yes % of soft/hard bottoms: global scale (LTAP):

% of coarse habitats: reef-scale

ConclusionsUsing SPOT-LANDSAT images

for mapping, inventory and monitoring of reefs?

Monitoring: Not directlyChange detection generally not possibleCoarse-habitats level not generally a relevant parameterWater quality generally not directly available

But provide: Geophysical parameters (exposure, bathymetry, residence time, geomorphology) Habitat mapping to stratify monitoring and establish new sites Generalize species indicator at reef scale

Timing: once or variable (catastrophic event)