Geostatistics 3D
Geostatistics 3D
v. 10.7
Objectives
Explore the various 3D interpolation algorithms available in GMS, including IDW and kriging. Visualize
the results of interpolation though cross sections and isosurfaces.
1 Introduction ......................................................................................................................... 2
1.1 Getting Started ............................................................................................................. 2
2 Importing a Scatter Point Set ............................................................................................. 3
3 Displaying Data Colors ....................................................................................................... 4
4 Z Magnification ................................................................................................................... 4
5 Creating a Bounding Grid .................................................................................................. 5
6 Simple IDW Interpolation .................................................................................................. 6
7 Displaying Isosurfaces ........................................................................................................ 6
7.1 Interior Edge Removal ................................................................................................. 7
7.2 Specified Range............................................................................................................ 8
7.3 Using the Vertical Anisotropy Option .......................................................................... 8
8 Interpolation ........................................................................................................................ 9
8.1 IDW Interpolation with Gradient Planes ...................................................................... 9
8.2 IDW Interpolation with Quadratic Functions ............................................................. 10
8.3 Other Interpolation Schemes ...................................................................................... 11
9 Viewing the Plume with a Cross Section ......................................................................... 11
9.1 Using the Truncation Option ...................................................................................... 13
10 Setting up a Moving Cross Section Animation ............................................................... 14
10.1 Display Options .......................................................................................................... 14
10.2 Setting Up and Playing the Animation ....................................................................... 14
11 Setting up a Moving Isosurface Animation ..................................................................... 14
12 Conclusion.......................................................................................................................... 15
1 Introduction
2. If GMS is already running, select File | New to ensure that the program settings
are restored to their default state.
It is first necessary to import a 3D scatter point set. These are similar to 2D scatter point
sets except that each point has a z coordinate (elevation) in addition to xy coordinates. As
with the 2D scatter point set, one or more scalar datasets can be associated with each
scatter point set. These can represent values such as contaminant concentration, porosity,
hydraulic conductivity, and so on.
The 3D scatter point set that will be imported and used with this tutorial has previously
been exported to a text file using a spreadsheet application. The file was imported into
GMS using the Import Wizard, and then saved as a GMS project file. Refer to the
“Geostatistics – 2D” tutorial for details on using the Import Wizard.
To import the project, do as follows:
4. Click Open to import the project and close the Open dialog.
Notice that the points are arranged in vertical columns (Figure 1). This hypothetical set
of points represents a set of measurements of contaminant concentration in the vicinity
of a leaky underground storage tank. Each column of points corresponds to a borehole or
the path of a penetrometer along which concentrations were measured at uniform
intervals. The goal of the tutorial is to use the tools for 3D geostatistics in GMS to
interpolate from the scatter points to a grid and generate a graphical representation of the
plume.
Next, it is necessary to change the display options so that the color of each point is
representative of the concentration at that point.
4. Click the Options… button to the right of Contours to bring up the Dataset
Contour Options – 3D Scatter Points – c dialog.
5. At the bottom left of the dialog, turn on Legend.
4 Z Magnification
It is necessary to magnify the z coordinate so that the vertical variation in the data is
more apparent.
To generate a graphical representation of the contaminant plume, first create a grid that
bounds the scatter point set. It will then be possible to interpolate the data from the
scatter points to the grid nodes. The grid will be used to generate isosurfaces.
Do the following to create the grid:
1. In the Project Explorer, expand all of the entries.
An interpolation scheme needs to be selected next. First use the inverse distance
weighted interpolation scheme (IDW).
13. Click OK to accept the defaults and close the Interpolate → Object dialog.
7 Displaying Isosurfaces
Now that interpolation has been completed on the nodes of the 3D grid, there are several
ways to visualize the contaminant plume. One of the most effective ways is to use
isosurfaces. Isosurfaces are the three-dimensional equivalent of contour lines
representing a surface of a constant value (contaminant concentration, in this case).
To define and display isosurfaces, do the following:
5. Click the Options… button to the right of Isosurfaces to bring up the Isosurface
Options dialog.
6. On the first row of the table, enter “3000.0” in the Upper Value column.
7. On the second row, check the box in the Fill between column.
8. In the section on the right, turn on Isosurface faces.
A series of edges are draped over the isosurface plot. These edges represent the
intersection of the isosurface with the grid cells. The edges are displayed to help
visualize the spatial variation or relief in the isosurface. However, it is sometimes useful
to inhibit the display of the edges in some areas. For example, in the regions where the
plume intersects the grid, the isosurface is flat. It is advisable to turn off the display of
the edges in this area since they provide little benefit.
1. Select the “ 3D Grid Data” folder in the Project Explorer to make it active.
Notice that the shell of the isosurface is all one color, but the interior of the isosurface
(where the isosurface intersects the grid boundary) varies in color according to the
contaminant concentration. To show more distinct color variation, do the following:
The scatter points being used were obtained along vertical traces. This means the
distances between scatter points along the vertical traces are significantly smaller than
the distances between scatter points along the horizontal plane. This disparity in scaling
causes clustering and can be a source of poor results in some interpolation methods.
The effects of clustering along vertical traces can be minimized using vertical anisotropy.
The z coordinate of each of the scatter points is multiplied by one over the vertical
anisotropy parameter prior to interpolation. Thus, if the vertical anisotropy parameter is
less than 1.0, scatter points along the same vertical axis appear farther apart, and scatter
points in the same horizontal plane appear closer together.
As a result, points in the same horizontal plane are given a higher relative weight than
points along the z axis. This can result in improved accuracy, especially in cases where
the horizontal correlation between scatter points is expected to be greater than the
vertical correlation (e.g., horizontal layering of soils, spreading of the plume on the top
of the water table).
To change the vertical anisotropy, do the following:
1. Select the “ 3D Scatter Data” folder in the Project Explorer to make it active.
8 Interpolation
2. In the Interpolation method section, click the Options… button to the right of
the Inverse distance weighted option to bring up the 3D IDW Interpolation
Options dialog.
3. In the Nodal function section, select the Gradient plane option.
7. Click OK to accept the defaults and close the Interpolate → Object dialog.
The Graphics Window will appear similar to Figure 7.
2. In the Interpolation method section, click the Options… button to the right of
Inverse distance weighted to bring up the 3D IDW Interpolation Options dialog.
3. In the Nodal function section, select Quadratic.
4. In the Computation of nodal function coefficients section, select Use all points.
8. Click OK to accept the defaults and close the Interpolate → Object dialog.
The Graphics Window will appear similar to Figure 8. Notice how the contour of the
isosurface layers has changed slightly from that in the previous section.
Two other 3D interpolation schemes (natural neighbor interpolation and kriging) are
supported in GMS. However, these schemes will not be reviewed in this tutorial. Feel
free to experiment with these techniques at a later time.
While isosurfaces are effective for displaying contaminant plumes, it is often useful to
use color-shaded cross sections to illustrate the variation in the contaminant
concentration.
To do this, first cut a horizontal cross section through the center of the plume:
3. Using the Create Cross Section tool, cut a horizontal cross section through
the grid and the middle of the isosurface by clicking to the left of the grid,
moving the cursor to the right of the grid, and double-clicking.
11. Click the Options… button to the right of Contours to bring up the Dataset
Contour Options – 3D Grid – c_idw_quad dialog.
12. In the Contour method section, select “Color fill” from the drop-down.
Notice the range of contaminant concentration values shown in the color legend at the
upper left corner of the Graphics Window (Figure 9). A large percentage of the values
are negative. This occurs due to the fact that a higher order nodal function was used.
Both the quadratic and the gradient plane nodal functions infer trends in the data and try
to preserve those trends.
In some regions of the grid, the values at the scatter points are decreasing when moving
away from the center of the plume. This decreasing trend is preserved by the
interpolation scheme; moreover, the interpolated values approach zero and eventually
become negative in some areas.
However, a negative concentration does not make sense. This problem can be avoided by
truncating the values in order to force all negative values to have a value of zero.
It is possible to create several cross sections at different locations in the grid to illustrate
the spatial variation of the plume. This process can be automated using the animation
utility in GMS. An animation can be generated showing a color-shaded cross section
moving through the grid.
Before setting up the animation, delete the existing cross section, turn off the color
legend, and reset the contour range.
2. Using the Select Cross Sections tool, select the cross section by clicking on
it, then press the Delete key.
2. In the Options section, turn on Cross sections / Isosurfaces and click Next to go
to the Cross Sections / Isosurfaces page of the Animation Wizard dialog.
3. Turn on the Animate cutting plane over specified XYZ range section.
4. Turn on Z.
Another effective way to visualize the plume model is to generate an animation showing
a series of isosurfaces corresponding to different isovalues.
2. In the Options section, turn on Cross sections / Isosurfaces and click Next to go
to the Cross Sections / Isosurfaces page of the Animation Wizard dialog.
3. Turn off the Animate cutting plane over XYZ range section.
4. Turn on the Animate isosurface over specified data range section.
5. Enter “1000.0” for the Begin value.
6. Enter “15000.0” for the End value.
7. Turn on Cap above and Display values.
12 Conclusion
This concludes the “Geostatistics – 3D” tutorial. The key concepts discussed and
demonstrated in this tutorial include the following:
Mesh-centered grids are better than cell-centered grids if using interpolation and
not using MODFLOW.
Vertical anisotropy can be used to help overcome the problem of grouping that is
common with data collected from boreholes.