Environmental Engineering Reference
In-Depth Information
arguments against any simplistic aggregation of areas “affected,” “modii ed,” “trans-
formed,” or “impacted” by human actions. And these categorical complications are
not the only challenge in assessing the aggregate consequences of anthropogenic
changes: whereas quantifying their large-scale extent has become easier thanks to
modern remote sensing techniques, major uncertainties remain. The global monitor-
ing of these changes became possible only with the launching of the Earth observa-
tion satellites and with the gradually improving resolution and multispectral images
produced by their sensors (the sequence was discussed in the i rst chapter). At the
same time, those who produce global maps and data sets of changing land use do
not do enough to stress many of the inherent limitations of the data, and hence the
untutored users of these products have unrealistic opinions about their accuracy,
reliability, and comparability.
What these limitations mean is perhaps best illustrated by taking a closer look
at the global satellite monitoring of cropland. Most of today's gridded global land-
use databases have a resolution of 5 arc-minutes, encompassing an area of about
9.2
9.2 km, or roughly 8,500 ha of farmland. Even on Canadian prairies with
their large holdings averaging 400-500 ha, that resolution would aggregate all crops
planted on about 20 different farms into a single data point. In most parts of Asia,
such a resolution would homogenize hundreds of different farms planted to scores
of different crops—and unlike on the Canadian prairies, with their single crop per
year, nearly all of those farms would harvest at least two crops a year, many i elds
would be triple-cropped, and suburban vegetable i elds would produce four to six
crops every year. Obviously, a 5-minute resolution is good enough to identify large-
scale patterns of land use but not to provide accurate assessments of specii c plant
composition or productivity, especially in multicropped areas.
Data series for the more recent presatellite eras must be assembled from the best
available national statistics or periodic land-use mappings. These sources use non-
uniform dei nitions of land-cover and land-use categories, their reliability varies
from excellent to dubious, and while they may offer enough quantitative pegs to
make reasonably good continent-wide estimates for nineteenth-century Europe or
North America, they have little or no information for pre-1900 Africa and large
parts of Asia. But even such questionable, fragmentary. and widely spaced numbers
are largely absent for the Middle Ages and entirely unavailable for antiquity: land-
cover reconstructions for those eras must rely heavily on assumptions based on
anecdotal written evidence and, where available, on painstaking paleoecological
reconstructions.
×
Search WWH ::




Custom Search