Michael F. Goodchild, Ashton M. Shortridge, & Peter Fohl
NCGIA, Dept. of Geography,
University of California Santa Barbara good@geog.ucsb.edu, ashton@geog.ucsb.edu,
fohl@geog.ucsb.edu
Traditionally, spatial data producers have used data specifications and summary statistics to report data quality information, while the onus of applying this quality information to account for spatial uncertainty has rested with the spatial data user. Here we demonstrate a conceptual framework to encapsulate a stochastic simulation algorithm with a geospatial data set. This approach allows the user to automatically generate a set of equally probable realizations of the spatial phenomenon using the best information available. Responsibility for providing the simulation method and specific model parameters rests with the data producer.
View the Java simulation machine, or read read on for more information.
This application demonstrates how the data producer can provide the
uncertainty model and simulation method to the data user, allowing the
user to identify the fitness of the data for any particular spatial application.It
also illustrates the simulation/propagation approach to modeling spatial
data uncertainty.
| In this example application, a quadrilateral parcel is defined by four
surveyed points. The observed data indicate that the parcel forms a square
100m.on a side, with an area of 10,000 square meters.
However, the surveyed points are subject to positional uncertainty. The magnitude of this uncertainty is characterized by a gaussian distribution with a mean of zero and a standard deviation of 2 meters. The diagram on the right indicates the nature both of the Research Question: What is the standard deviation associated with the area of the land parcel, given the positional uncertainty information? |
 |
Back to Ashton Shortridge's
home page.
Back to NCGIA-Santa
Barbara page.