Page 1 of 17
Civil Engineering Seminar
13 October 2006
GEOSTATISTICS:WHAT IS IT AND WHAT’S IT
GOOD FOR?
Donald E. Myers
Emeritus Professor of Mathematics
Emeritus Joint Professor of Hydrology
Member of Committee on Remote Sensing and Spatial Analysis
NOTE: you may need to copy/paste some of these hyperlinks as opposed
to merely clicking on them, in particular if the URL is longer than one line
http://www.u.arizona.edu/~donaldm
COMMERCIAL
I will be teaching Geostatistics next semester, Mathematics/Geography
574, TTH 12:30-1:45
WHAT IS GEOSTATISTICS?
Geostatistics is concerned with spatial data and uncertainty
Each data value is associated with a location
Each data value/location has a support
Changing the support can be useful but also cause problems
The data locations need not be on a regular pattern.
Spatial data is often expensive to collect and thus there usually isn’t
enough of it.
Page 2 of 17
Collecting it may alter the environment, e.g drilling wells or boreholes
Uncertainty can occur or appear in a problem/application in at least
two ways: (1) something has not yet occurred and might occur in
different ways, (2) lack of knowledge, i.e., data. Geostatistics is more
concerned with the latter than with the former.
Some examples of spatial data
Geohydrological parameters
Hydraulic conductivity
Porosity
Water table/aquifer properties
Depth to groundwater (head)
Wolfcamp data
www.uni-muenster.de/ZIV/Mitarbeiter/BennoSueselbeck/s-html/hel
pfiles/aquifer.html
http://www.ic.arizona.edu/ic/math574/analyses/wolfcamp/wo_plt.gif
Concentrations of various natural substances in the subsurface
Ore grades
Soil nutrients
Contaminant concentrations
In an aquifer
In the soil
Weather/climate characteristics (could be spatial-temporal as well)
Humidity
Temperature
Page 3 of 17
Wind velocity
Components of wind direction
An Important Characteristic Commonly Observed for Spatial Data
If two locations are close together the values are more “similar” than
if the locations are far apart.
A key question of course is, what do we m