This presentation given in cooperation by CIFOR and the Earth Institute focuses on fires in tropical regions: what's influencing them, what can help reduce them and which monitoring methods can be applied.
UiPath Community: Communication Mining from Zero to Hero
Science and tools for fire adaptation and mitigation in tropical landscapes
1. Science and tools for fire adaptation and mitigation
CIFOR-EARTH INSTITUTE PARTNERSHIP
Miguel Pinedo-Vasquez, Victor Gutierrez-Velez, Katia
Fernandes, Christine Padoch, Maria Uriarte, Walter Baethgen and
Ruth DeFries
2. Outline
1. Many tropical areas are experiencing profound and rapid
landscape change.
2. Some of these changes are resulting in an increased risk
of agricultural fires escaping.
3. Drivers of increased fire incidence are complex and
interrelated, often in non-linear ways
4. We are developing science and tools that may help
communities both mitigate and adapt to climate change
and fire risk in tropical landscapes.
4. Fires in humid tropical regions
• Fire is not a natural phenomenon but
its use can be dated to 8,000 yrs BP.
(Bush et al. 2007)
• It is the cheapest tool for land
preparation, pasture and plantation
management (Fernandes et al. 2011)
• Where climate change leads to
decreased precipitation
(Malhi, 2008), fire uses need to adapt
5. Multiple dimensions of change influencing fire risk
Land cover dimensions
Demographic
Shifts
Absentee
Landlords
FIRE RISK
Land use/cover
changes
Urbanization
Mobility
Large pastures
plantations
Droughts
Social/demographic
dimensions
Variability
Seasonality
Dry
spells
Climatic dimensions
Vegetation
change
6. Fires are increasing in areas with declining rural
Prop. change in
population in Western Amazonia
rural population
More
people
Fewer
people
Uriarte et al. (2012).
7. Gutierrez-Velez et al (submitted)
x SPI
SPI***
x SPI
SPI***
SPI***
x SPI
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
palm***
SPI***
SPI***
c. palm
alm***
SPI***
0.0
More fire
0.5
0.0 0.0
palm***
c. palm
palm***
1.0
0.5 0.5
dary***
alm***
c. palm
1.5
1.0 1.0
owth***
dary***
palm***
2.0
1.5 1.5
rass***
owth***
dary***
2.0 2.0
hort***
rass***
owth***
hort***
rass***
hort***
Standardized parameter
estimate
Secondary forests and adult oil palm plantations can help
OccurNoRdNoYoungerX
OccurNoRdNoYoungerX
OccurNoRdNoYoungerX
reduce fire occurrence.
-0.5 -0.5 -0.5
-1.0 -1.0 -1.0
8. Gutierrez-Velez et al (submitted)
SPI***
SPI***
x SPI
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
SPI***
palm***
SPI***
SPI***
c. palm
palm***
SPI***
8
x SPI
SPI***
x SPI
0.0
More fire
0.5
0.0 0.0
palm***
c. palm
palm***
1.0
0.5 0.5
dary***
palm***
c. palm
1.5
1.0 1.0
owth***
dary***
palm***
2.0
1.5 1.5
rass***
owth***
dary***
2.0 2.0
hort***
rass***
owth***
hort***
rass***
hort***
Standardized parameter
estimate
Drought severity reduces the ability of secondary forests but not of
OccurNoRdNoYoungerX
OccurNoRdNoYoungerX
OccurNoRdNoYoungerX
adult oil palm plantations to reduce fire occurrence
-0.5 -0.5 -0.5
-1.0 -1.0 -1.0
9. Fire Early Warning System in Western Amazonia
J/A/S Western Amazonia fire anomalies can be forecast from retrospective Sea
Surface Temperature.
3.00
Precip-JAS
2.00
1.00
0.00
-1.00
-2.00
Fernandes, K., et al. (2011).
Fires-JAS
10. Fires Early Warning System in Western Amazonia
JAS 2010 Fire-Anomalies predicted since (a) April, (b) May and (c) June with 95%
confidence.
(a)
Fernandes, K., et al. (2011).
Fernandes, K., et al. (2011).
(b)
(c)
11. We are developing science and tools for local
adaptation to fire risk as well as for mitigation:
1. Designing early warning systems for adaptation to
climate variability.
2. Providing institutional support for fire prevention and
intervention systems.
3. Mapping vulnerability to fire based on sociodemographic dynamics and landscape change.
4. Designing land cover management strategies to
mitigate climate and fire risk in transitional landscapes
In many areas, smallholder landscapes that involved a finely-grained mosaic of forests, fallows, fields, settlements, and other features is being changed. Mosaics are being replaced with far larger patcheswhether they be larger pastures, plantations of commodity crops, or protected areas.
These changes also reflect a number of changes in
Censuses from 1987, 1993, 2003, 2007
We developed satellite products and incorporated them into a Bayesian statistical model to assess the effect of land cover changes and drought severity on fire activity in the study area. The parameter estimates for each variable indicate the sign and strength of the correlation between predictors and fire occurrence (the probability of each pixel to burn). Positive values mean more fire. Land covers representing different stages of regrowth portray an U shape trend with pastures and falllow promoting fire and degraded pastures and secondary forests inhibiting it. Oil palm plantations exhibit a negative linear trend with age class. Young plantations (0-5 yr) increase fire probability, adolescent (5-10 yr) have an non-significant effect and adult plantations (>10 yr) reduce it.
Drought is the variable with the highest correlation. Parameter estimates for multiplicative interactions between land covers and drought exhibit a positive trend. LCs representing more advanced stages of forest regrowth increase fire probability as it becomes dryer. In contrast adult plantations do not promote fire even during anomalously dry years. Results show that land cover management can help reducing fire occurrence.
Climate change mitigation in the context of droughts and rainforests can work through fire prevention. Early warning systems (EWS) for droughts and fires are available at IRI, for both Western Amazon and in Kalimantan-Indonesia. Dry season (July-August-September) fire and precipitation anomalies averaged over the Western Amazon domain vary close together (top plot). JAS precipitation is also driven by North Tropical Atlantic SST anomalies (bottom plot), so if the climate (SST) is the main driver of fire variability, we should be able to predict anomalies in upcoming fire seasons using seasonal forecast.Fire anomalies prediction for western Amazon is shown for JAS 2010. The colorbar shows standardized anomalies (no unit), so if you really need to mention what the map is showing is: “standardized anomalies of fire count at each pixel”. The fire data used is “Active Fires from MODIS”.The warm colors indicate prediction of an active fire season made in April, May and June. The dots basically show the pixels where we got it right. White areas in the map represent, for most part, areas where fires are very rare and are not included in the prediction.The seasonal fire prediction can be used, a few months prior to the beginning of the dry season, to identify regions where concentrated effort to prevent fires should be directed.
Fire anomalies prediction for western Amazon is shown for JAS 2010. The colorbar shows standardized anomalies (no unit), so if you really need to mention what the map is showing is: “standardized anomalies of fire count at each pixel”. The fire data used is “Active Fires from MODIS”.The warm colors indicate prediction of an active fire season made in April, May and June. The dots basically show the pixels where we got it right. White areas in the map represent, for most part, areas where fires are very rare and are not included in the prediction.The seasonal fire prediction can be used, a few months prior to the beginning of the dry season, to identify regions where concentrated effort to prevent fires should be directed.
The link for our product is too long, so I am copying it here instead of having it in the slide. You can email it if necessary.http://iridl.ldeo.columbia.edu/home/.katia/.FirePrediction/.Jul/.JAS_SST_Forecast/figviewer.html?my.help=&map.T.plotvalue=2005.0&map.lat.units=degree_north&map.lat.plotlast=5.025N&map.url=lon+lat+fig-+colors+thin+states+thinnish+countries+-fig&map.domain=+%7B+%2FJAS_SST_Forecast+-2+2+plotrange+%2FT+2005.0+plotvalue+lat+-20.075001+5.0250001+plotrange+%7D&map.domainparam=+%2Fplotaxislength+432+psdef+%2Fplotborder+72+psdef+%2FXOVY+null+psdef&map.zoom=Zoom&redraw.x=20&redraw.y=14&map.lat.plotfirst=20.075S&map.lon.plotfirst=81.55W&map.lon.units=degree_east&map.lon.modulus=360&map.lon.plotlast=65.45W&map.JAS_SST_Forecast.plotfirst=-3&map.JAS_SST_Forecast.units=Standard+Deviation&map.JAS_SST_Forecast.plotlast=3&map.newurl.grid0=lon&map.newurl.grid1=lat&map.newurl.land=countries&map.newurl.plot=colors&map.plotaxislength=432&map.plotborder=72&map.fnt=Helvetica&map.fntsze=12&map.color_smoothing=1&map.XOVY=auto&map.iftime=25&map.mftime=25&map.fftime=200