The proposed research will focus on land-cover change, its quantification and forcing
functions, and on natural forest dynamics in the State of Acre, Brazil. Acre is one of the
prime frontiers for evaluating current trends of deforestation and logging. This region
already has large-scale colonization projects, logging activities, and extensive cattle
ranching, and has a population growing at 3%/year, one of the highest rates in Amazonia.
In addition, the region has unique characteristics for comparative studies of
degradation with extensive areas of eutrophic soils intermingled with nutrient-poor
oxisols and ultisols. The area has some of the most innovative alternative land-uses in
Amazonia -1.4 million ha of extractive reserves - that maintain forest ecosystem function
and provide income to forest residents. Acre also has about half of the 18 million ha of
bamboo-dominated forests of western Amazonia. The expansion and contraction of these
low-biomass forests have strong implications for the carbon dynamics of the region.
Uncertainties in the rate of deforestation and in the biomass of primary forests currently
limit estimates of anthropogenic modifications of the Amazonian carbon budget. Rates of
deforestation based on similar data sets differed by a factor of 2 in Acre during the late
1980s and estimates of biomass for primary and secondary forests in Amazonia vary by 30%
We will use multi-temporal satellite imagery extensive and ground truthing to determine
the rates, patterns, and causes of land-cover change during the past decade, at various
scales, concentrating in eastern Acre, and will compare this with other sites on the
western LBA transect. Geo-referenced permanent plots in different forest types will be
used to determine forest biomass, growth and mortality rates, as well as serve as
calibration sites for new satellites. Fire frequency and distribution will be determined
with ground truthing to verify AVHRR hot pixel data. We plan to participate in parallel
studies at primary and secondary tower sites along the LBA western transect. In
collaboration with other researchers, we will analyze the fire susceptibility of different
forest types, including secondary forests. We will use high-resolution remote sensing to
map logging activities, natural clearings, and forest types. Multi-temporal imagery will
be used to estimate the dynamics of the bamboo-dominated forests in the region.
These detailed studies will be compared with the regional data produced by other
collaborators to estimate precision and accuracy of basin-wide deforestation and burning
rates. These data will be stratified as a function of land tenure (spontaneous and planned
settlements, ranching, logging and extractive reserves) and soil nutrient status. We will
evaluate the patterns of land-cover change, incorporating socio-economic and policy
modifications, such as the new zoning programs and reduction in legal area for
deforestation that may drastically alter future scenarios of land-use in the region. The
knowledge generated will be incorporated into the Zoning Program for the State of Acre and
the Ecosystem Management and Global Change Programs of the Federal University of Acre.
This proposal seeks to address the following questions, derived from the LBA-ECO
1. Given that estimates of state-level deforestation rates differ in some cases by a
factor of two, how may we improve the accuracy of remote sensing estimates of
deforestation and land-use at the local and regional scales such that the data are more
useful both for modeling carbon fluxes and for land-use management?
2. The official data on logging and burning are frequently misleading. How can more
dependable estimates of these activities be made? How can we differentiate the burning of
felled forests, pastures and undergrowth in logged and primary forests?
3. Lateral migration of forest types with differing biomass, as well as in situ changes
in mortality and growth, affect carbon fluxes between the atmosphere and the biosphere.
How important are such migration and in situ changes for the carbon budget in western
4. How may the knowledge generated by this and other NASA research reach local and
regional societies so as to promote the prospect of sustainable land-use?
The following hypotheses form the core of the study effort:
- Hypothesis 1. Scale and classification factors seriously affect deforestation estimates.
- Hypothesis 2. Small-scale deforestation and burning are poorly measured on annual basis.
- Hypothesis 3. Hot pixels of NOAA/AVHRR imagery indicate fire frequency on a daily basis
and distribution in western Amazonia.
- Hypothesis 4. Coupling deforestation rates with forest types will substantially improve
estimates of carbon fluxes in western Amazonia.
- Hypothesis 5. The life cycles of bamboo-dominated forests affect the carbon budget of
western Amazonia on a decade time-scale .
- Hypothesis 6. Carbon cycling of large trees in primary forests is spatially and
temporally discontinuous. Forest C uptake is nearly uniform spatially and continuous in
time, while release is discrete in time and space.
- Hypothesis 7. Annual Landsat TM imagery improves estimates of logging activities in
- Hypothesis 8. Re-growing forests represent a minor sink of carbon in western Amazonia.
- Hypothesis 9. Agroforestry systems in converted lands increase vegetative biomass and
decrease fire frequency as compared to neighboring converted lands without agroforestry
Last Updated: May 15, 1998