Séminaire : Quels outils pour un changement d'échelle dans la gestion des insectes d’intérêt économique ?
Remote sensing for spatial ecology
Agnès BEGUE (CIRAD, UMR TETIS)
Atelier CIRAD Oct 2011 Agnès BEGUE • 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…
2 The remote sensing offer
3 A more than 150-year old technique !
Boston, à partir d’un ballon 1860, J. W. Black
1859 Invention of photography (1839) - First known aerial photo (Tournachon - “Nadar”, France) - First known saved aerial photo (1860 - James Wallace Black.) 1ère image CORONA - USSR 18 août 1960
1960s First meteorological and military satellites : Nimbus (1964) / Corona (1960)
4 Satellite remote sensing (1/2)
Satellites privés THRS 18 16 14 12 10 Satellite SPOT 8 Satellite Landsat 6 4
2 Nombre satellites lancés Nombre satellites 0 1972 1977 1982 1987 1992 1997 2002 2007 2012 Année de lancement
In the optical domain :http://gdsc.nlr.nl/gdsc/information/earth_observation/satellite_database (janv. 2010)
5 Satellite remote sensing (2/2)
1000
100
10
Multispectral 1 Panchromatique
Log(Résolution spatiale (m)) (m)) Log(Résolutionspatiale Hyperspectral Thermique 0.1 1960 1970 1980 1990 2000 2010 2020 Année lancement
Optical domain :http://gdsc.nlr.nl/gdsc/information/earth_observation/satellite_database (janv. 2010)
6 How to choose a satellite image (which images for which application )?
Study site Image size
Objects/classes to identify Spatial resolution
Surface parameters to quantify Spectral band
Time period and frequency Archive/Programming (tasking)
Budget Image cost
Partnership Image licence
Technical skills / ancillary data Image level
7 How to choose a satellite image (which images for which application )?
Study site Image size
Objects/classes to identify Spatial resolution
Surface parameters to quantify Spectral band
Time period and frequency Archive/Programming (tasking)
Budget Image cost
Partnership Image licence
Technical skills / ancillary data Image level
8 VEGETATION (3200 km)
• Echelle régionale (106 km2) – SPOT/VEGETATION
– MODIS QuickBird (15 km)
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) Spatial resolution
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.
10 Spatial resolution vs Image size
Modis
40
30 Landsat
20 Aster SPOT 10 Formosat Ikonos IRS QuickBird
0 Photo aérienne Résolution spatiale (m) 0 50 100 150 200 Taille image (km)
11 How to choose a satellite image (which images for which application )?
Study site Image size
Objects/classes to identify Spatial resolution
Surface parameters to quantify Spectral band
Time period and frequency Archive/Programming (tasking)
Budget Image cost
Partnership Image licence
Technical skills / ancillary data Image level
12 Spectral bands (1/2)
THERMAL INFRARED
VISIBLE
13 Spectral bands (2/2)
Surface Spectral band parameters Biomass, Leaf area, Visible + Near Infrared (large spectral bands) vegetation cover … = multi-spectral
Plant N content, Visible + Near Infrared (narrow spectral bands) soil organic matter, = Super – hyper-spectral soil components …
Evapo-transpiration Thermal Infrared Urban temperature … Soil moisture Micro-waves = radar Surface roughness … MNT images Tree height, DEM… Radar altimetry DSM MNT terrain naturel
14 How to choose a satellite image (which images for which application )?
Study site Image size
Objects/classes to identify Spatial resolution
Surface parameters to quantify Spectral band
Time period and frequency Archive/Programming (tasking)
Budget Image cost
Partnership Image licence
Technical skills / ancillary data Image level
15 Archive or tasking
• 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…)
16 Acquisition frequency
• 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 …
Ground Track after 1 Day Ground Track after 7 Days 17 How to choose a satellite image (which images for which application )?
Study site Image size
Objects/classes to identify Spatial resolution
Surface parameters to quantify Spectral band
Time period and frequency Archive/Programming (tasking)
Budget Image cost
Partnership Image licence
Technical skills / ancillary data Image level
18 Image cost
Cost = f(Archive/tasking,Commande resolution, min image en Euros size, pre-processing level…)
25 Quickbird
15 IKONOS
5 SPOT
Landsat
Coût (Euros/km²) (Euros/km²) Coût Coût -5 5 15 25 35 -5
Résolution (m)
Satellite image cost (tasking) (the size of the circle is proportional to the minimum order in €)
19 Aerial remote sensing (1/5)
Ultra-light aircraft
20 Aerial remote sensing (2/5)
VISIBLE PROCHE INRAROUGE INDICE DE VEGETATION
Site de La Mare, le 19 avril 2006 RED-EDGE (ROUGE/PIR) INFRAROUGE THERMIQUE
V. Lebourgeois et al. (2006) 21 Aerial remote sensing (3/5)
Sugarcane
V. Lebourgeois et al. (2006) 22 Aerial remote sensing (4/5)
Detection of weeds
V. Lebourgeois et al. (2006) 23 Aerial remote sensing (5/5)
Precision farming
Characterizing the INSIDE PLOT HETEROGENEITY
V. Lebourgeois et al. (2006)
24 The remote sensing information content
25 Spectral information
26 Textural Information B
B QuikBird Panchromatic 0.6m B
SC
Source : G. Lainé, CIRAD
Natural Sugarcane Banana vegetation trees
27 Structural information
Source : Quickbird P Mechanized banana field
Source : Quickbird MS+P Not mechanized banana field
28 Source : Gérard Lainé, CIRAD Organisation level
Domaine Domaine forestier agricole
Cultures et Boisement Boisement Reboisement lâche dense terrains nus
Ombre feuillages Sol nu
TRES HAUTE RESOLUTION C. Puech (2001) 29 Temporal information (seasonal)
192618102162621181912142517 11 913juillet4décembreoctobrejanvierfévrier juillet septembre avrilaoûtjuinmaimarsseptembremai 20032004 20032004 2002 20032002 200420022003 Paysage agricole 2003 20022004 Ile de La Réunion
nuages sol nu -
Activité chlorophylienne
+
Bégué et al. (2009) 30 6 km Temporal information (seasonal)
repousse Plantation 2003 Plantation 2004 800
600
400
200 Indice de végétation (1-1000) 0 avr-02 oct-02 mars-03 sept-03 mars-04 sept-04
31 Temporal information (annual)
Bruzzone (2003)
32 Temporal information (annual)
Bruzzone (2003)
33 Temporal information (mid/long term trend)
Jong et al., 2011
Bruzzone (2003)Bruzzone (2003)
34 Image pre-processing
Not to be underestimated !!!
35 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.
36