This proposal addresses a key need to accurately monitor
land cover change at repeated intervals throughout Amazônia.
Spatially-explicit data on land cover dynamics in response to land use
change are essential to extrapolate process-level understanding of carbon,
nutrient, and trace gas fluxes over the larger region.
Currently, monitoring efforts to quantify land cover change based on
analysis of Landsat data are time-consuming, laborious, and data-intensive.
Estimates from different analyses often yield conflicting results,
highlighting differences in land cover definitions and methodologies applied by
We propose to collaborate with colleagues from the
Brazilian National Institute for Space Research (INPE) to develop efficient,
accurate, and repeatable methods for monitoring land cover dynamics throughout
the region. The emphasis will be on
integrating data from different sensors at multiple resolutions to apply methods
that are as automated and practicable as possible.
Specifically, we propose to:
1) develop and test a nested approach using coarse
resolution (250-500m) MODIS data to identify locations undergoing land cover
change and Landsat data for more detailed characterization and analysis of those
identified locations. Very high
resolution (1-4m) IKONOS data and in situ
observations will be used to validate the results. Through a nested approach, we aim to reduce the amount of
high resolution data requiring time consuming analysis without sacrificing
wall-to-wall coverage of the entire region.
2) examine an alternative approach to describe land cover
dynamics in terms of subpixel percentage tree cover to circumvent the
definitional differences in terms such as “forest”, “nonforest”,
“degraded forest”, “deforestation”, and “regrowth.”
Through such an approach, we aim to improve the ability to identify
subpixel changes in tree cover and provide estimates of forest area and change
that are consistent in space and time.
3) Establish collaborative relationships and exchanges of
methods and data between INPE, the University of Maryland, and the LBA-ECO team.
4) Provide subsets of 250m and 500m MODIS data and improved
estimates of land cover dynamics throughout the region to the LBA community
through the University of Maryland Global Land Cover Facility (already linked to
the LBA data distribution system).