Identifying degraded forests in an Amazonian landscape from
remote-sensing
C. BOURGOIN, L. BLANC, J. FERREIRA, V. GOND,
N. BAGHDADI, A. HASSANALI, F. LAURENT, S. LE CLEC’H, J. OSZWALD, I. TRITSH
International Congress for Conservation Biology August 6, 2015, Montpellier
Context & definitons - Objectives - Field work - Multisource remote sensing - Perspectives
Deforested lands = 20% Brazilian Amazonian forest (INPE, FAO 2013)
2
How to identify and characterize a range of forest degradation within this fragmented landscape ?
Combining : field work &
multisource remote sensing
Context & definitons - Objectives - Field work - Multisource remote sensing - Perspectives
Paragominas, Pará, Brazil
25 m Emergents 35 m
F1
25 m 35 m
F2
25 m 35 m 15 m
F3
25 m15 m
F4
15 m 6 m
15 m 6 m
Forest degradation typology
Context & definitons - Objectives - Field work - Multisource remote sensing - Perspectives
F1
F2
F3
F4
4 Context & definitons - Objectives - Field work - Multisource remote sensing - Perspectives
Optical data
RANDOM FOREST :
AGB regression and prediction
RADAR data Time series data
LANDSAT 8 SPOT 5 ALOS-1 SENTINEL-1 MODIS
100km
Berenguer E., Ferreira J., Gardner T., Barlow J.
25m 250m
“A Large-Scale Field Assessment of Carbon
Stocks in Human-Modified Tropical Forests.” (2014)
AGB plots
Field measured AGB (Mg/ha)
Predicted AGB (Mg/ha)
RMSE = 2.1339 Mg/ha Variables importance
AGB PREDICTION
Context & definitons - Objectives - Field work - Multisource remote sensing - Perspectives
Remote sensing
REMOTE SENSING
UNDERSTANDING
MODELISATION
I
Definitions Field typology
TERRITORIAL GOVERNANCE
GI S DEC IS
O
ON T
LO S
THANK YOU
FOR YOUR ATTENTION !
Contact : bourgoin.clement2@gmail.com