Creating a Scattergram of Observed v Computed Node Values

RockWorks | Utilities | Solid |
Statistics | Krajewskigram

Use this program to create an XY plot or scattergram of observed G values versus computed node G values for corresponding locations in an existing solid model.  This is one method of determining the accuracy of the solid model.

! This tool requires that the source data, from which the solid model was computed, be displayed in the main Utilities datasheet so that source data can be compared to the modeled data.

Menu Options
Step-by-Step Summary
 

Menu Options

• Input Columns:  
  • X (Easting), Y (Northing): Select the columns in the main data sheet that contain the X and Y location coordinates for the source data.
    These can be Eastings and Northings in meters or feet, local coordinates, etc. See Defining your Datasheet Coordinates for more information.
  • Z (Elevation): Select the data sheet column that contains the elevation (Z) values for the source data.
  • G (Grade):  Select the data sheet column that contains the original measurements that are represented in the solid model, below. These source G values will be plotted along the X axis, against the computed node values. 
  
  
• Solid Model: Click on this item to select the name of the existing solid model (.RwMod) file, created from the data now displayed in the main data sheet, whose node values are to be displayed against the source G values in the scatter plot. 
  
  
• Symbols: Click on the sample to select the desired symbol style and color.
• Symbol Size: Click to establish the size of the symbols, as a percent of diagram width. 
• Border Annotation: Insert a check here to activate the plotting of titles, ticks, and labels along the horizontal and vertical axes. Expand this item to access the settings. 
  • Left/West Axis, Right/East Axis, Basal/South, Top/North Axis: Insert a check in any of these check-boxes to activate their plotting. Expand each item to access common settings: 
    • Ticks: To turn tick marks on for any activated axis, insert a check in the appropriate check-box. Expand the Ticks to set size (as a percent of the diagram width) and color. 
    • Labels: To turn labeling on for any activated axis, place a check in the appropriate check-box. Expand this heading to set label angle, size (as a percent of the diagram width), color, and font type.  
    • Titles: To turn titles on for any activated axis, place a check in the appropriate check-box. Expand Titles to enter the text to be plotted for the title, and to set label angle, size (as a percent of the diagram width), color, and font type.
    • Subdividers: Expand this to turn on/off the plotting of a horizontal and/or vertical grid of lines, at major and minor intervals. Once activated, expand the items to choose the colors and line styles for the selected grids. 
      
      
• Manual Ranges: Insert a check in this box if you want to establish a range of data for the diagram that is either less than the actual data values (to filter the data), or greater than the actual data values (to extend the axes for comparing different data sets).  If the Manual Ranges check-box is left blank, the program will use all scanned data and dimension the X and Y axes to accommodate those data.  Expand this item to enter the desired ranges.  
  • X Minimum, X Maximum: Click on these items to type in the minimum and maximum data values to be represented for the X axis (observed values).. 
  • Y Minimum, Y Maximum: Click on these items to type in the minimum and maximum data values to be represented for the Y axis (solid model node values).
• Manual Subdivision: Insert a check here to override the program-computed axis labeling intervals. Expand this item to access settings. 
  • X Minor: Type in the interval for the X axis at which small tick marks are to be plotted. 
  • X Major: Enter the X axis interval at which major tick marks and labels are to be plotted. This must be an even multiple of the X Minor setting. 
  • Y Minor: Type in the interval for the Y axis at which small tick marks are to be plotted. 
  • Y Major: Enter the Y axis interval at which major tick marks and labels are to be plotted. This must be an even multiple of the Y Minor setting. 
    
    
• Diagram Size: Expand this item to enter the height and width of the diagram. These values would typically represent inches, however note that you can scale the diagram both larger and smaller for both screen display and printer output in the RockPlot2D plotting window. 
   

Step-by-Step Summary

1. Access the RockWorks Utilities program tab. 
2. If necessary, open the XYZG data from which the solid model was created.  This is necessary because the program needs to know the original observed values against which the solid model node values will be compared.
   • If the model was originally created in the Utilities, the XYZG data was probably stored in an .RwDat file:  Use the Project Manager to open this file.
   • If the model was originally created in the Borehole Manager, use the Borehole Manager's File | Export | datatype | Utilities Datasheet options to transfer source data from the Borehole Manager to the Utilities datasheet for use in this program.
3. Select the Solid | Statistics | Krajewskigram menu option. 
4. Enter the requested menu settings, described above.
5. Click the Process button to continue. 

The program will read the original data values from the main data sheet. It will determine the calculated solid model G-values for each control point location and create an X,Y scattergram of observed versus calculated G-values. 

It will also compute the correlation coefficient and list the result within the diagram title (assuming that you have enabled the border annotation).  This number ranges between zero (no correlation) to 1.0 (perfect correlation). 

Note: The correlation coefficient provides a quantitative comparison between the observed data and the predicted model.  It does not reflect how reasonable the model is.  In other words, a model with a high correlation coefficient isn't necessarily good.  Generally, as the model resolution increases, the correlation coefficient will also increase, up to a point.  This point, where increase in node density doesn't provide a significant increase correlation, could be considered the point of diminishing returns, where the increased processing time, model size, and modeling artifacts aren't worth the small increase in correlation. 

6. View / save / manipulate / print / export the diagram in the RockPlot2D window. 
   
   

Back to Solid Menu Summary

RockWare home page