Nature, food, and water
of data-source agendas, ephemerality of some data, and especially the scarcity
of independent peer review needed for dependable scholarship).
Large detailed satellite images (30 m cell size) of urban regions were printed.
Number, type, distance, form, and area of objects were directly measured with
ruler and planimeter on the images by a single observer. In almost all cases
each type of measurement was made on all 38 urban regions within a few
days. Spatial measurements for patterns in this chapter required an average of
20 minutes per urban region, and those for Chapter7 required 35 min. To evalu-
ate observer consistency or variability, periodically five repeated measurements
of the same attribute were made. Compared with the first measurement, the
average of the five measurements varied on average 3 % (range 0--9 %), suggesting
that this methodology produced reasonable consistency.
The degree of variability and resulting confidence in data and measurements
can be further understood from the following examples. Population data for
urban regions were for slightly different years (1999 to 2002) and probably dif-
fered in quality and area coverage (Chapter5).Perimeter and surface area mea-
surements of the metropolitan areas depend on determining cut off points for
house-lot density and width of greenspace, plus their consistent application.
Perhaps, on average, 5 % of a metro-area perimeter was near the cut-off points,
with the percentage ranging from about 0--10 % over the set of regions. Attributes
such as rivers, lakes, reservoirs, airports, shipping/ferry ports, and large mine
sites were clear on the images, so confidence in the associated numbers was
veryhigh. Some biodiversity sites and recreation/tourism sites, for example, were
clear on an image, but many were not, and thus had to be marked in approx-
imate locations. In addition, considerable interpretation was necessary in esti-
mating whether certain sites had the appropriate attributes (rare species or rare
natural community, significant recreation or tourism from the city in a single
day), in locating a site on the image, and in deciding whether the site was a
major one to be included. In another case, urban region boundaries, which were
invisible on the images, were mainly delineated using six variables (Chapter5),
with different variables being primary in different portions of a boundary.
On the other hand, many measurements varied negligibly, such as number of
marked objects and distance between two points. These examples attempt to
provide insight into the degree of confidence appropriate for different portions
of the extensive information base underlying the color maps and their spatial
analyses.
In essence, most of the patterns and results highlighted in these two chapters
were based ondata and measurements with the high degree of confidence quite
normal in science. In cases where more variability existed in the measurements