HR: 11:25h
AN: H22B-05 [Abstracts]
TI: Mapping Global Urban Extent and Intensity for Environmental Monitoring and Modeling
AU: * Schneider, A
EM: ams@icess.ucsb.edu
AF: University of California Santa Barbara Department of Geography and Institute for
Computational Earth System Science, Ellison Hall 5703, Santa Barbara, CA 93106-4060, United States
AU: Friedl, M A
EM: friedl@bu.edu
AF: Boston University Department of Geography and Center for Remote Sensing, 675
Commonwealth Avenue, Boston, MA 02215, United States
AB:
The human dimensions of global environmental change have received increased attention in policy, decision-
making, research, and even the media. However, the influence of urban areas in global change processes is still
often assumed to be negligible. Although local environmental conditions such as the urban heat island effect
are well-documented, little or no work has focused on cross-scale interactions, or the ways in which local urban
processes cumulatively impact global changes. Given the rapid rates of rural-urban migration, economic
development and urban spatial expansion, it is becoming increasingly clear that the ‘ecological footprint' of cities
may play a critical role in environmental changes at regional and global scales.
Our understanding of the cumulative impacts of urban areas on natural systems has been limited foremost by a
lack of reliable, accurate data on current urban form and extent at the global scale. The data sets that have
emerged to fill this gap (LandScan, GRUMP, nighttime lights) suffer from a number of limitations that prevent
widespread use. Building on our early efforts with MODIS data, our current work focuses on: (1) completing a
new, validated map of global urban extent; and (2) developing methods to estimate the subpixel fraction of
impervious surface, vegetation, and other land cover types within urbanized areas using coarse resolution
satellite imagery. For the first task, a technique called boosting is used to improve classification accuracy and
provides a means to integrate 500 m resolution MODIS data with ancillary data sources. For the second task, we
present an approach for estimating percent cover that relies on continuous training data for a full range of city
types. These exemplars are used as inputs to fuzzy neural network and regression tree algorithms to predict
fractional amounts of land cover types with increased accuracy. Preliminary results for a global sample of 100
cities (which vary in population size, level of economic development, and spatial extent) show good agreement
with the expected morphology in each region.
DE: 1803 Anthropogenic effects (4802, 4902)
DE: 1834 Human impacts
DE: 1843 Land/atmosphere interactions (1218, 1631, 3322)
DE: 1855 Remote sensing (1640)
SC: Hydrology [H]
MN: 2007 Joint Assembly