Remote sensing for spatial ecology - Cirad · 0 Photo aérienne 10 20 30 40 0 50 100 150 200 Taille...

Séminaire : Quels outils pour un changement d'échelle

dans la gestion des insectes d’intérêt économique ?

Atelier CIRAD Oct 2011 Agnès BEGUE

Remote sensing for spatial ecology

Agnès BEGUE (CIRAD, UMR TETIS)

2

• Many references on remote sensing for spatial ecology

– « Remote sensing » and « ecology » (121)

– « Remote sensing » and « habitat » (108)

– « Remote sensing » and « biodiversity » (37)

– « Remote sensing » and « pest » (8)

– « Remote sensing » and « insect » (7)

– …

• Applications in spatial ecology

– Land cover classification (qualitative RS) and spatial analysis

– Land surface parameters (quantitative RS) and modeling

– Land surface change (change detection and trend analysis)

• Remote sensing offer

– Satellite remote sensing / Aerial remote sensing

• Remote sensing information content

– Spectral, spatial, temporal dimensions…

3

The remote sensing offer

4

A more than 150-year old technique !

Invention of photography (1839)- First known aerial photo (Tournachon -“Nadar”, France) - First known saved aerial photo (1860 -James Wallace Black.)

1859

Boston, à partir d’un ballon

1860, J. W. Black

First meteorological and military satellites : Nimbus (1964) / Corona (1960)

1960s

1ère image CORONA - USSR 18 août

1960

5

0

2

4

6

8

10

12

14

16

18

1972 1977 1982 1987 1992 1997 2002 2007 2012

Année de lancement

Nom

bre

sate

llite

s lancés

Satellite remote sensing (1/2)

Satellites privés THRS

Satellite SPOT

Satellite Landsat

In the optical domain :http://gdsc.nlr.nl/gdsc/information/earth_observation/satellite_database (janv. 2010)

6

Satellite remote sensing (2/2)

Optical domain :http://gdsc.nlr.nl/gdsc/information/earth_observation/satellite_database (janv. 2010)

0.1

1

10

100

1000

1960 1970 1980 1990 2000 2010 2020

Année lancement

Lo

g(R

és

olu

tio

n s

pa

tia

le (

m))

Multispectral

Panchromatique

Hyperspectral

Thermique

7

How to choose a satellite image

(which images for which application )?

Partnership Image licence

Study site Image size

Objects/classes to identify Spatial resolution

Time period and frequency Archive/Programming (tasking)

Budget Image cost

Technical skills / ancillary data Image level

Surface parameters to quantify Spectral band

8







Budget Image cost



VEGETATION (3200 km)

QuickBird (15 km)

• Echelle régionale (106 km2) – SPOT/VEGETATION

– MODIS

LAN DSAT (180 km) SPOT (60 km)

• Echelle locale (10 -> 104 km2) – SPOT/LANDSAT (103 / 104 km2) – QuickBird/Ikonos (centaine km2) – Photos aériennes (dizaine km2)

10

SPOT XS = 20m Ikonos MS = 4m Ikonos P = 1m

10 palm trees 1-2 palm trees <1 palm tree

For a thematic question,

the best spatial resolution is not always the finest.

Spatial resolution

11

Landsat

SPOTFormosatIRS

Aster

Ikonos

QuickBird

Photo aérienne0

10

20

30

40

0 50 100 150 200

Taille image (km)

Réso

luti

on

sp

ati

ale

(m

)Modis

Spatial resolution vs Image size

12







Budget Image cost



13

Spectral bands (1/2)

VISIBLE

THERMAL INFRARED

http://en.wikipedia.org/wiki/Image:Human-Visible.jpg

http://upload.wikimedia.org/wikipedia/commons/4/44/Human-Infrared.jpg

14

Surface

parameters

Spectral band

Biomass, Leaf area,

vegetation cover …

Visible + Near Infrared (large spectral bands)

= multi-spectral

Plant N content,

soil organic matter,

soil components …

Visible + Near Infrared (narrow spectral bands)

= Super – hyper-spectral

Evapo-transpiration

Urban temperature

…

Thermal Infrared

Soil moisture

Surface roughness

…

Micro-waves = radar

Tree height, DEM…

DSM

Radar altimetryMNT terrain naturelMNT terrain naturel

MNT imagesMNT images

Spectral bands (2/2)

15







Budget Image cost



16

• Archives :

• Landsat (1972), SPOT (1986), SPOT5 (2002)

• QuickBird (2001), Ikonos (1999)

• NOAA (1982), VEGETATION (1998), MODIS (1999)

• Aerial photos …

• Tasking :

• Only some satellites are programmable : SPOT, THRS

• Cost >

• Not garanteed (tasking conflicts, clouds…)

Archive or tasking

17Ground Track after 1 DayGround Track after 7 Days

• The acquisition frequency depends on :

– Satellite orbital parameters + Target latitude

– Sensor field of view + Sensor depointing capacities

• Different time scales according to the process :

– Daily monitoring (low resolution satellites)

• Natural hazards, water stress …

– Seasonnal monitoring (low and high resolution satellites)

• Primary production, fraction of soil covered by vegetation…

– Annual monitoring (high and very high resolutions satellites)

• Change in land use/ land cover …

Acquisition frequency

18







Budget Image cost



19

Satellite image cost (tasking)(the size of the circle is proportional to the minimum order in €)

Commande min en Euros

-5

5

15

25

-5 5 15 25 35

Résolution (m)

Co

ût

(E

uro

s/km

²)

Landsat

SPOT

IKONOS

Quickbird

Commande min en Euros

-5

5

15

25

-5 5 15 25 35

Résolution (m)

Co

ût

(E

uro

s/km

²)

Landsat

SPOT

IKONOS

Quickbird

Image cost

Cost = f(Archive/tasking, resolution, image size, pre-processing level…)

20

Aerial remote sensing (1/5)

Ultra-light aircraft

21

VISIBLE PROCHE INRAROUGE

RED-EDGE (ROUGE/PIR) INFRAROUGE THERMIQUE

INDICE DE VEGETATION

Site de La Mare, le 19 avril 2006

V. Lebourgeois et al. (2006)


22

Sugarcane



23

Detection of weeds



24

Characterizing the

INSIDE PLOT

HETEROGENEITY

Precision farming



25

The remote sensing information content

26

Spectral information

27

SC

B

B

B

Textural Information

Source : G. Lainé, CIRAD

Sugarcane Banana

trees

QuikBird Panchromatic 0.6m

Natural

vegetation

28

Structural information

Mechanized banana fieldSource : Quickbird P

Not mechanized banana fieldSource : Quickbird MS+P

Source : Gérard Lainé, CIRAD

29

Organisation level

Ombre feuillages Sol nu

Domaine

forestier

Domaine

agricole

TRES HAUTE RESOLUTIONC. Puech (2001)

Boisement

lâche

Boisement

denseReboisement

Cultures et

terrains nus

30

-

+

nuages

sol nu

Activité

chlorophylienne

12 juin 20026 juillet 200214 août 20029 septembre 2002

18 octobre 200210 janvier 200326 février 200325 mars 200326 avril 20034 mai 200317 juin 200321 juillet 200321 août 200319 décembre 2003

13 mai 200418 juin 200419 août 200411 septembre 2004

26 octobre 2004

6 km

Paysage agricole

Ile de La Réunion

Temporal information (seasonal)

Bégué et al. (2009)

31

0

200

400

600

800

avr-02 oct-02 mars-03 sept-03 mars-04 sept-04

Ind

ice

de

vé

gé

tatio

n (

1-1

00

0)

repousse

Plantation 2003

Plantation 2004

Temporal information (seasonal)

32

Temporal information (annual)

Bruzzone (2003)

33

Temporal information (annual)

Bruzzone (2003)

34

Temporal information (mid/long term trend)

Bruzzone (2003)Bruzzone (2003)

Jong et al., 2011

35

Image pre-processing

Not to be underestimated !!!

36

Conclusions

• A very large offer in terms of spatial data (satellite and/or

aerial images) : resolution, image size, spectral bands,

repetitivity, length of time series…

– Increasing number of Very High Resolution satellites;

– New satellite concepts (daily visit in High Resolution);

– A « democratisation » of the image (Google Earth…)

– For the « democratisation » of the costs… be patient…

• Most of the remote sensing applications are of interest for

ecology :

– Qualitative and quantitative description of the main landscape

components (vegetation, soil, water, altiutude…), and their respective

spatial distributions.

Remote sensing for spatial ecology - Cirad · 0 Photo aérienne 10 20 30 40 0 50 100 150 200 Taille...

Documents

Transcript of Remote sensing for spatial ecology - Cirad · 0 Photo aérienne 10 20 30 40 0 50 100 150 200 Taille...