Recent trends in fires and land cover change in Western Indonesia

1
Recent trends in fires and land cover change in
Western Indonesia

• Douglas O. Fuller
• Department of Geography and Regional Studies
• University of Miami, Florida
• Collaborators T.C. Jessup, Agus Salim, Erik
  Meijaard, Martin Hardiono

2
Talk Outline

• Background Drivers, Ecology, Consequences
• The ENSO-fire relationship Understanding
  climate and human actions
• Carbon Emissions, Peat swamp forest, and REDD
• Projecting the future with land change models

3
Carbon Emissions

• 1.2 Gt yr-1 (12 percent of the global total) from
  tropical deforestation and forest degradation
  (Nature Geoscience 2009)
• Pan et al. (Science 2011) report a global forest
  sink of 1.1 Gt yr-1
• 0.3 Gt yr-1 from tropical peat fires, mostly in
  Indonesia (4 percent of global total)
• Emissions from Indonesia (Sumatra and Borneo)
  estimated at 30x during the 2006 El Nino vs. the
  2000 La Nina

4
Study area
5
Vegetation Cover
6
Social, Economic, and Cultural Consequences of
Fire

• 1997 fires cost Indonesia 1 billion in lost
  tourism, transportation and health impacts.
• Rampant land conversion implicated in the loss of
  cultural diversity
• Region-wide effects haze spreads over much of
  Southeast Asia

7
The haze dilemma
8
GHG Emissions
Indonesia forest cover 95-120 million ha, 2-4
percent annual deforestation rate.
Peat fires important in Indonesia 1997-98 fires
emitted 40 percent CO2 from fossil fuels. Source
Page et al. Nature, 2002
9
Process of Land-cover Change in Kalimantan,
Indonesia
Selective Logging
Disturbed Lowland Forest
Agricultural burning
Alang-alang savanna
10
The fire transition

• A new theory that accounts for anthropogenic
  changes in tropical fire regimes through time
• Fires are rare in closed forests except during
  exceptional climatic events (extreme El Nino for
  example)
• Fires become more seasonal as forests are
  converted and remaining high biomass needs to be
  removed quickly to make way for plantations
• As permanent crops are established and land
  values rise, fires diminish as people practice
  fire suppression to protect valued assets

11
Some Satellite Systems for Mapping Fire
Diurnal Patterns of Burning and Satellite Overpass
Fuller, 2000, Prog. Phys. Geogr.
12
Landsat TM image showing industrial plantations
13
Fire vs. ENSO indices
Fuller Murphy, 2006, Clim Change
14
Fire-SOI The influence of land-cover type
Non-forest (agriculture, degraded land, pasture)
Swamp and mangrove forest
Tropical moist forest
r 0.75
Fuller Murphy, 2006, Clim Change
15
Annual Time Scale
Fuller Murphy, 2006, Clim Change
16
Extending the Time Series Using MODIS Fire
Overlap
17
Fuller Meijaard, 2010, submitted
18
St
TS Models and Decomposition Xt St Rt et ?
additive model Xt St x Rt x et ? multiplicative
model
Rt
Xt
et
19
Cross-correlations between fire and ENSO,
2001-2010
ALL-M PSF-A LOW-A MONT-A P/S-M O/M-M
NINO12-M -0.28(-37) -0.21(-45) 0.25(-9) -0.19(-2) 0.24(6) 0.22(-11) 0.21(-11) 0.25(-9) 0.33(41) 0.30(3) 0.19(-11) 0.19(-13) 0.29(22) 0.32(34) 0.28(6) 0.20(23) -0.22(27) 0.28(23) 0.42(41) 0.20(-23) 0.24(-37) 0.18(12) -0.24(-37) 0.16(-17) 0.24(6) -0.30(-37) -0.23(-45) 0.26(-8) -0.22(-6) 0.20(46)
NINO3-M 0.40(-10) -0.24(-37) 0.48(-8) 0.18(-7) 0.32(-11) 0.42(-11) 0.24(4) 0.46(-8) 0.17(26) 0.45(-12) 0.38(-11) 0.22(-3) 0.46(-9) 0.33(8) 0.40(-12) 0.23(-7) 0.17(6) 0.31(-6) 0.18(-37) 0.22(-10) 0.39(-11) -0.16(-37) 0.42(-8) 0.17(-7) 0.35(-12) 0.38(-10) -0.27(-38) 0.48(-8) 0.21(8) 0.24(-11)
NINO4-M 0.39(-10) 0.35(-3) 0.40(-10) 0.29(-3) 0.38(-11) 0.34(-11) 0.30(-21) 0.36(-6) 0.28(-10) 0.47(-14) 0.31(-11) 0.27(-21) 0.41(-9) 0.25(-10) 0.41(-11) 0.22(-5) 0.19(-21) 0.28(-4) 0.23(-18) 0.26(-8) 0.39(-10) 0.31(-3) 0.41(-10) 0.26(-2) 0.38(-11) 0.37(-10) 0.33(-3) 0.39(-10) 0.27(-3) 0.33(-10)
NINO3.4-M 0.41(-10) 0.33(-8) 0.47(-8) 0.16(1) 0.39(-12) 0.41(-12) 0.34(0) 0.44(-7) 0.25(1) 0.50(-14) 0.37(-12) 0.30(-9) 0.47(-9) 0.23(1) 0.45(-12) 0.25(-6) 0.34(2) 0.33(-5) 0.27(2) 0.26(-10) 0.40(-10) 0.29(-4) 0.43(-8) 0.17(31) 0.40(-12) 0.39(-9) 0.31(2) 0.45(-8) 0.15(1) 0.32(-12)
Black whole series, red 2001-2006, blue
2007-2010 (May)
Fuller Meijaard, 2010, submitted
20
Evidence consistent with the decoupling
hypothesis

• Maximum cross-correlations decreased across the
  two time segments (except for PSF)
• Time lags between fires and ENSO increased
  noticeably
• 3) Seasonality increased in certain transitional
  land
• cover types (especially fire-susceptible
  forests)

21
Peat carbon 2 billion pledged to help
Indonesia implement REDD
Soros wants to turn Indonesia into a pilot
project for his carbon trading plan.
22
(No Transcript)
23
Evidence of change from Landsat
24
Some background on peat deposits

• About 55 percent of PSF have been logged and
  drained, which exposes peat surfaces that burn
  readily during droughts (seasonal or otherwise)
• Range in age from 2-26 Kyr
• Range in thickness from 1-20 meters
• Contain up to 18x the carbon of the above-ground
  biomass
• Total carbon store of 55 (/-10) Gt in Indonesia
• Largest deposits in Central Kalimantan
• When drained, they subside due to oxidation (60
  percent) and shrinkage (40 percent)

25
Hooijer et al., 2010, Biogeosciences
26
Peat depths from core samples Jaenicke et al.
(2009), Geoderma
27
Change in carbon stocks
Cconversion Si(CAFTERi - CBEFOREi ) ?A TO
OTHERSi ? gross emissions where Cconversion
change in carbon stocks on land converted to
another land category, t C yr-1 CAFTERi
carbon stocks on land type i immediately after
the conversion, t C ha-1 CBEFOREi carbon
stocks on land type i before the conversion, t C
ha-1 ?A TO OTHERSi area of land use i
converted to another land use category in a
certain year, ha yr-1 i type of land use
converted to another land use category.
Source IPCC, 2006, IPCC Guidelines for National
Greenhouse Gas Inventories.
28
More to the point.how REDD is supposed to work
29
Ministry of Forestry Maps
Hutan Rawa 2005
Hutan Rawa circa 1995
Hutan Rawa MoF map 2006
3,505,425 ha of Hutan Rawa
2,660,692 ha of Hutan Rawa
Both maps derived from interpretation of Landsat
imagery
30
Research Design
31
Local roads
Rivers
Fires 1997
Reforestation
Deforestation 1995-2005
Fires 2005
32
Constrained 3x3
1.39 million ha lost
GEOMOD - 2020
2005
0.9 million ha lost
Dinamica EGO - 2020
0.8 million ha lost
LCM - 2020
Fuller et al., 2011, Environmental Management
33
reforestation/regeneration (RR) between 2005-2010
and protection of Sebangau NP
48,000 ha of regrowth through replanting or
natural regeneration
34
BAU vs. Some Regeneration
2020 BAU (no PSF regeneration Considered) 1.86
million ha
2020 Regeneration scenario 2.28 million ha
35
Forest Loss Projections
Fuller et al., 2011, Env. Mgmt
36
Conclusions

• LUCC models are useful to explore possible
  outcomes given a range of scenarios
• Our results indicate that Indonesia can meet
  between 36-81 percent of its 2020 target for
  reduced greenhouse gas emissions of 0.78 Gt CO2
  equivalent (e) by implementing peatland
  restoration and other REDD interventions in
  Central Kalimantan.

37
Research Frontiers

• Results reflect emissions from deforestation only
  not degradation (RED not REDD)
• Fluxes from oxidizing peat not well known, so
  emissions baselines are difficult to establish
• More accurate accounting will include degradation
  and carbon sequestration (Gtnet)
• Extend fire analysis to continue testing fire
  transition theory using cross-border comparisons

38
THANK YOU!