Any organization or individual who needs to statistically explore
data and create surfaces for a number of variables will benefit from this statistical
software package. Some of the various fields that use Geostatistical Analyst
include agriculture, geology, meteorology, hydrology, archaeology, forestry,
oceanography, fishery, health care, and environmental studies.
Geostatistical Analyst complements Spatial
Analyst. Most of the interpolation methods available in Spatial Analyst
are represented in Geostatistical Analyst as well, but in Geostatistical
Analyst there are many more statistical models and tools and all their parameters
can be manipulated to derive optimum surfaces. Additionally, Geostatistical
Analyst provides exploratory spatial data analysis tools not available in
Spatial Analyst. Spatial Analyst has many functions in other areas, such
as map algebra, combinational operators, and data conversion.
Where
ArcGIS Spatial Analyst includes rudimentary interpolation methods, ArcGIS Geostatistical
Analyst expands the number of deterministic and geostatistical interpolation
methods and provides many additional options. In particular, Geostatistical
Analyst provides a variety of different output surfaces such as prediction,
probability, quantile, and error of predictions. Surfaces can be displayed
as grids, contours, filled contours, and hillshades or any combination of these
renderings. These surfaces can be exported in raster and shapefile formats
for working together with other extensions such as ArcGIS Spatial Analyst.
Geostatistical Analyst also includes an interactive set of exploratory spatial
data analysis tools for exploring the distribution of the data, identifying
local and global outliers, looking for global trends, and understanding
spatial dependence in the data.
Introduction to Modeling Spatial Processes Using Geostatistical
Analyst[PDF-2.13 MB] introduces geostatistical theory and the tools implemented in Geostatistical
Analyst. Case studies provide examples of statistical
analysis of environmental data using ArcGIS Geostatistical Analyst. Educational
and research papers provide articles on various aspects of geostatistical theory
and applications. The manual for Geostatistical Analyst discusses usage of
methods implemented in the software. Shop
online for this manual and a variety of geostatistics books for advanced
users.
More advanced geostatistical textbooks include
Cressie, N. 1993. Statistics for Spatial Data, rev. ed. Wiley-Interscience.
Chiles, J., and P. Delfiner 1999. Geostatistics. Modeling Spatial Uncertainty.
Wiley-Interscience.
Waller, L., and C. Gotway 2004. Applied Spatial Statistics for Public Health Data. Wiley-Interscience.
Any data that has associated spatial coordinates can be used in
Geostatistical Analyst. This data can be arrayed spatially as random points,
as a regular grid, or as centroids of polygons. Examples are temperature measured
at monitoring stations, DEMs, and cancer rates per county.
Exploratory spatial data analysis is a set of graphical tools for determining statistical data features and which interpolation method is appropriate for the data. With it you can explore the distribution of the data, look for global and local outliers, look for global trends, examine spatial autocorrelation, and understand the correlation between multiple data sets. The views in exploratory spatial data analysis are interactive with ArcMap. Data selected with these tools will also be selected in ArcMap and in all of the other exploratory tools.
Interpolation methods derive surfaces from measured samples to predict values for each location in a landscape. ArcGIS Geostatistical Analyst provides two groups of interpolation methods: deterministic and geostatistical. All methods rely on the similarity of nearby sample points to create the surface. Deterministic methods use predefined mathematical functions for interpolation. Geostatistical methods rely on statistical features of the data. Geostatistical models also assess the uncertainty of the predictions.
Yes. One of the goals of exploratory spatial data analysis is to
find unusually large and small values (outliers), which can be either errors
or the most interesting data in the data set. Semivariogram/Covariance Cloud
and Voronoi Map tools are especially useful for finding unusual data.
By default, the Semivariogram/Covariance tools work with data sets having a maximum
of 300 data points, totaling nearly 45,000 point pairs. This value was chosen
for performance and for usability reasons: It is difficult to find all the interesting
data among a huge number of points. However, you may increase the maximum setting,
C:\Program Files\ArcGIS\Utilities\AdvancedArcMapSettings.exe, as shown.
If you do increase the maximum, you may experience slow performance with large data sets.
ArcGIS Geostatistical Analyst will work with as few as 10 data
measurements. However, the more data measurements you have, the better your
prediction is likely to be. Data with weak spatial correlation usually requires
more measurements than data with strong spatial correlation. For kriging, the software requires a minimum of 10 data points to create
a surface.
No. Geostatistical Analyst provides the necessary tools for data
exploration and variography analysis. Many analytical tools are included to
create accurate surfaces such as detrending, declustering, checking for bivariate
normality, data transformations, cross validation, validation, and model comparison.
The semivariogram model used in Geostatistical Analyst depends on the properties of the data. The spherical semivariogram is used by default because it is the most popular in both geostatistical literature and software. From a theoretical point of view, the best model is a J-Bessel model.
We suggest modifying lag size and number of lags such that the distance of significant
correlation (range) occupies about two-thirds of the x-axis.
It is always good to compare cross validation statistics for several semivariograms
and choose those that produce smaller prediction errors, see Introduction to
Modeling Spatial Processes Using Geostatistical Analyst.
To determine the best interpolation technique, use exploratory spatial data analysis tools. For example, based on the result of trend analysis, you may want to use the local polynomial deterministic interpolation method to remove large-scale variation from the data before using one of the kriging models.
As a rule, deterministic interpolation techniques (inverse distance weighted,
radial basis functions, and local polynomial interpolation) should not be used
for decision making, because they do not provide information on how good their
predictions are. Geostatistical interpolation techniques (e.g., kriging) can
be chosen based on the result of exploratory spatial data analysis and diagnostics
(cross validation and validation).
Show/Hide All Answers
