Image Processing Reference
large blocks of GDP per capita below US$1,000 per person. This is an artifact
of the mapping method and Guatemala's highly unequal distribution of land and
wealth (Gini coefficient for land is 0.85, with 1 being perfect inequality) (Todaro
1994 ). The Pacific coast area is dominated by large cattle, coffee, sugar cane,
and cotton estates. Population density on these estates is low most of the year.
These areas do see, however, a large influx of seasonal workers to harvest crops,
especially coffee. The specific dynamics of land use and ownership are not well
represented by Fig. 17.9 , but overall patterns are verifiable by an intimate
knowledge of the country. We must temper this new method of mapping poverty
and wealth in the developing world (where census data are often unavailable or
unreliable) with the caution that it should only be conducted along with exten-
sive ground truthing that can explain anomalies like the case illustrates above.
Nonetheless, we believe that further research into the meaning and potential of
this idea of mapping wealth and poverty with population and nighttime imagery
should be explored.
Nighttime satellite imagery derived from the DMSP OLS has potential for
monitoring and measuring many anthropogenic phenomena at a global scale.
While nighttime satellite imagery is no substitute for an on the ground census
of the population it can be used in innovative and interesting ways to supple-
ment mapping human presence and activity on the earth. The case study
exploring exurbia in the United States shows how the nighttime imagery may
provide information that is not captured by finer resolution imagery such as
Landsat. The case study exploring applications of DMSP OLS imagery in
Guatemala demonstrates some of the potential and pitfalls of using this kind
of imagery in developing countries. Many countries of the world lack the
financial and/or institutional resources to conduct useful censuses. Models
derived from readily available satellite imagery that have been validated in
parts of the world where good ground based information is available could
serve as reasonable proxy measures in countries that lack such information.
The LandScan data product takes advantage of many of these ideas. In addi-
tion, if a country has some limited resources with which to conduct an incom-
plete census of its population, existing imagery for that country could be used
to help design statistical sampling strategies for a limited census. These sam-
pling strategies could be designed to maximize the effectiveness and accuracy
of proxy measures derived from satellite imagery. Future nighttime image
products from NPOES will have finer spatial and spectral resolution which
are likely to improve the number and quality of applications to which this
kind of imagery can be used.