ENVISAT/ASAR data for forest observations in Siberia

1
ENVISAT/ASAR data for forest observations in
Siberia

• Thuy Le Toan, Julien Lhermitte
• CESBIO, Toulouse, France
• Andreas Wiesman, Urs Wegmuller
• Gamma Remote Sensing

2
SIBERIA-II Multi-Sensor Concepts for Greenhouse
Gas Accounting of Northern Eurasia
CONTEXT AND STUDY REGION
3
CONTEXT AND OBJECTIVE

• Need of forest and biomass map for the
  estimation of
• Carbon budget
• In SIBERIA-I project, a 1 M km2 map has been
• produced using Tandem ERS InSAR and JERS
• Although limited to biomass less than 50t/ha,
  the
• Siberia-I map was found essential to quantify
  Siberian
• forests as Carbon sources or sinks
• For SIBERIA-II region (3 M km2), ENVISAT ASAR
  data
• are investigated

4
STUDY OBJECTIVE

• Investigate the use of ASAR for forest and
  biomass mapping
• ASAR AP (multi-dates, polarisation, incidence)
• INSAR (Narrow Swath, Image mode) in Winter
• ASAR WideSwath

5
ASAR Repeat Pass Interferometry

• Background
• Tandem ERS interferometric coherence is related
  to
• forest biomass
• C-band (ERS and ASAR) 35 day repeat pass
  coherence was
• found not exploitable for biomass retrieval,
  except in certain
• favorable conditions

• Questions
• Is the Siberian winter provide sufficiently
  stable conditions for exploitable coherence?
• Relations with forest biomass ?

6
ASAR Interferometric pairs, winter 2004
7
Analysis of coherence
488 m
330 m

• Higher coherence
• In Northern regions (gt60N)
• . For smaller baselines (lt500 m)

597 m
Increasing baseline
736 m
1000 m
65 N
56N
8
Temperature data
During January-February, temperature in Southern
regionscan have peaks close to zero--gtthawing-free
zing can occur Northern regions temperature well
below zero
Bolshemurstinsky (56N)
Tura (64N)
9
Northern regions Comparison with Landsat,
forest database not yet available for
quantitative analysis
Fire scar
Landsat ETM 22 June 2002
ASAR Tura site (Northern regions) R Coherence
29 Dec-8 March 2004 G Mean backscatter B
Backscatter ratio
10
Southern regions map from Siberia-I available
for comparison
Siberia-I
R Coherence, Bratsk site, 21 Jan-25 Feb G Mean
backscatter B Backscatter ratio
11
Analysis of coherence as a function of Siberia-I
classes Forest scene
12
Analysis of coherence as a function of Siberia-I
classes Scene with agriculture and forests
13
Analysis of coherence as a function of Siberia-I
classes Scene with low coherence (600 m Baseline)
14
Observations in Winter Coherence Decreasing from
forest classes 10 t/ha to gt50 t/ha as expected
Agriculture, bogs, grass, open forest
unexpected lower coherence due likely to
freeze/thaw during the 35 day period (Southern
region) Backscatter intensity Lower on frozen
bare fields, increasing with Siberia-I
class Backscatter change No significant change
1 Water 2 Agriculture, bogs, grass 3 Open forest,
cut,fire 4 Forest 10-30 t/ha 5 Forest 30-50
t/ha 6 Forest gt50 t/ha
15
Analysis and mapping of ASAR APP WS data
Approach 1. Analysis of SAR APP data using GIS
forest database - polarisation temporal
variation - relation with biomass 2.Determinatio
n of forest and biomass indicators 3.Mapping
procedure applied to WS data
16
30kmx30km
APP
GIS database (IIASA))
G HV RB HH
17
Polarisation and temporal behaviour of forest
biomass classes
High biomass

• large seasonal variation
• important ground effects at HH,
• HV and HH/HV sensitive to biomass
• (2.5 dB in Winter)

18
Polarisation and temporal behaviour of forest
biomass classes Intersite comparison
HV
HH/HV
19
Analysis results
Forest/Non forest and Biomass classifiers Possibl
e use of
HV HH/HV HV Winter/ HV Summer HV multidate
Winter or Summer VV in Winter VV multidate Summer
Wideswath
From the analysis results, it is possible to use
VV multitemporal change of WideSwath data to
produce large scale forest mapping in Siberia
20
Processing chain

• Developed using
• Gamma ASAR (Gamma RS)
• Multi-image filtering
• (Quegan et al., 2000)
• Temporal change
• (Le Toan et al., 1997)

21
ASAR WS Biomass map (2004)
Siberia-1 Biomass map (1997-8)
22
Forest and biomass map from ASAR WS data 400 km x
900 km
23
Summary on forest and biomass mapping

• Repeat pass InSAR coherence from ASAR data in
  Siberian
• Winter may be exploitable in Northern taiga
  (gt55), if
• baseline lt500m
• Polarisation diversity of APP is useful, but
  restricted to
• local mapping (100 km)
• Multitemporal WideSwath data (strip of 400 km
  swath)
• provide large scale mapping in Siberia
• Larger biomass range is expected to be retrieved
  from
• L-band SAR (ALOS)