This is a simple example of error propagation, or the effects of underlying data uncertainty on
the products of GIS analysis. Given the simple model used and the straightforward nature
of the calculation, it is perhaps surprising that current GIS software does not attempt to
estimate confidence limits on area estimates, or to truncate the digits of reported area
calculations to make them consistent with known accuracy. It is necessary to supply an
estimate of positional accuracy, but that is often readily available in metadata, or in the
positional tolerance values established for cleaning operations or for overlay.
What is perhaps more significant is that the achievable accuracy comes as a surprise, not only
to naive users of GIS but to those who are supposedly wise to the accuracy issue. A simple
sense of achievable accuracy in GIS operations should surely be part of every GIS education. We
currently devote time and attention to accuracy standards, and students are expected to know the
positional accuracies of standard topographic maps. But we do not currently insist on the
ability to translate such knowledge into its implications for basic GIS functions.
The experience of the past decade has shown that accuracy in geographic data is a substantial
problem. To address it will require the combined efforts of cartographers, GIS users,
geographers, and specialists with an understanding of the levels of accuracy present in each major
type of geographic data. The remainder of the paper discusses two approaches to the problem,
one methodological and one statistical. Because the approaches are rooted in different
paradigms, it is possible to pursue both of them simultaneously.
This is a simple example of error propagation, or the effects of underlying data uncertainty on the products of GIS analysis. Given the simple model used and the straightforward nature of the calculation, it is perhaps surprising that current GIS software does not attempt to estimate confidence limits on area estimates, or to truncate the digits of reported area calculations to make them consistent with known accuracy. It is necessary to supply an estimate of positional accuracy, but that is often readily available in metadata, or in the positional tolerance values established for cleaning operations or for overlay.What is perhaps more significant is that the achievable accuracy comes as a surprise, not only to naive users of GIS but to those who are supposedly wise to the accuracy issue. A simple sense of achievable accuracy in GIS operations should surely be part of every GIS education. We currently devote time and attention to accuracy standards, and students are expected to know the positional accuracies of standard topographic maps. But we do not currently insist on the ability to translate such knowledge into its implications for basic GIS functions.The experience of the past decade has shown that accuracy in geographic data is a substantial problem. To address it will require the combined efforts of cartographers, GIS users, geographers, and specialists with an understanding of the levels of accuracy present in each major type of geographic data. The remainder of the paper discusses two approaches to the problem, one methodological and one statistical. Because the approaches are rooted in different paradigms, it is possible to pursue both of them simultaneously.
การแปล กรุณารอสักครู่..
