BACKTR Process

To access this process:

  • Estimate ribbon >> Validate >> Geostatistical Processes >> From Normal.
  • BACKTR is accessed automatically using the Advanced Estimation wizard.
  • View the Find Command screen, select BACKTR and click Run.
  • Enter "BACKTR" into the Command Line and press <ENTER>.

See this process in the Command Table.

Note: This is a superprocess and running it may have an effect on other Datamine files in the project.

Process Overview

The process back-transforms data from a normal distribution. 

For details of the method refer to:

GSLIB.  Geostatistical Software Library and User's Guide by Clayton V.Deutsch and Andre G.Journel, published by Oxford University Press, Second Edition, 1998.  ISBN 0-19-510015-8.

There are some differences between Datamine and GSLIB conventions.  However, unless otherwise stated, Datamine conventions are used.  Where required these are converted to GSLIB conventions by the process. Please see the copyright notice at the bottom of this article.

The SGSIM process includes an option that carries out a normal transformation prior to the simulation.  However if the option is not used then the simulated values will be normally distributed, and BACKTR will be required to back transform them..

Program Limits

BACKTR is limited to 500000 samples in the input (IN) file.

The following additional limitations currently apply to the folder name of the project:

  • The full path name of the project folder must not include spaces, for example c:\database\consim\my project1 is not allowed.

System Files

The following log files are generated by the BACKTR process:

_bktlog.txt

Log file.  Only useful if there is a problem.

_bkt_*.txt

Temporary system files.  These will be deleted if the process terminates cleanly.   

_sp*.dm

Temporary Datamine files.  These will be deleted if the process terminates cleanly.
 

All files matching the template _bkt_*.txt and _sp*.dm will be deleted as the process terminates.  Therefore you should not use any of these file names.

Input Files

Name

Description

I/O Status

Required

Type

IN1

Input data file.  This must include the normally distributed field NORMVAL that is to be back transformed.

Input

Yes

Table

REFDIST

Input file containing the transformation lookup table.  The file must include the field ORIGREF , containing the original data values, and the field NORMREF defining the corresponding normal score values.

Input

Yes

Table

Output Files

Name

I/O Status

Required

Type

Description

OUT

Output

Yes

Block Model

Output file containing the back transformed values. This contains the same data as the IN file, but with the added back transformed data field BACKVAL .

Fields

Name

Description

Source

Required

Type

Default

NORMVAL

Field in the input data file IN defining the normal score values to be back transformed.

IN

Yes

Numeric

Undefined

ORIGREF

Field in the input REFDIST file defining the original data values.

REFDIST

Yes

Numeric

Undefined

NORMREF

Field in the input REFDIST file defining the normal score values.

REFDIST

Yes

Numeric

Undefined

BACKVAL

Field to be created in the output file OUT for the back transformed data values.  If not specified then the field BACKVAL will be created,

Undefined

No

Numeric

BACKVAL

Parameters

Name

Description

Required

Default

Range

Values

MINNORM

Minimum value of NORMVAL field in IN file to be used for back transformation.

No

-4

Undefined

Undefined

MAXNORM

Maximum value of NORMVAL field in IN file to be used for back transformation.

No

4

Undefined

Undefined

MINBACK

Minimum value of BACKVAL field to be created in OUT file.

No

0

Undefined

Undefined

MAXBACK

Maximum value of BACKVAL field to be created in OUT file.

No

+

Undefined

Undefined

LOTAIL

Back-transformation method in the lower tail of the distribution to a minimum grade of MINBACK .

OptionDescription
1,

Linear interpolation.

2,

Power model interpolation.  The power used is defined by LOPAR .

No

1

1,2

1,2

LOPAR

Power parameter used in back-transformation of grades in the lower tail of the distribution to a minimum of MINBACK .  LOTAIL must be set to 2.

No

1

0,+

Undefined

UPTAIL

Back-transformation method in the upper tail of the distribution to a maximum grade of MAXGRADE .

OptionDescription
1,

Linear interpolation.

2,

Power model interpolation.  The power used is defined by UPPAR .

4,

Hyperbolic model extrapolation using power parameter defined by UPPAR .

No

1

1,4

1,2,4

UPPAR

Power parameter used in back-transformation of grades in the upper tail of the distribution to a maximum of MAXGRADE .  UPTAIL must be set to 2 or 4.

No

1

0,+

Undefined

Example

!BACKTR &IN(nscore1),&REFDIST(nsctran1),&OUT(backtr1),*NORMVAL(NSCORE),
             *ORIGREF(V),*NORMREF(TRANDATA),*BACKVAL(V1),@MINNORM=-4.0,
             @MAXNORM=4.0,@MINBACK=0.0,@MAXBACK=2000.0,@LOTAIL=1.0,
             @LOPAR=1.0,@UPTAIL=1.0,@UPPAR=1.0
 

Copyright

The command BACKTR is based on the GSLIB program BACKTR.

Copyright (C) 1996,  The Board of Trustees of the Leland Stanford
Junior University.  All rights reserved.   
 
The programs in GSLIB are distributed in the hope that they will be useful, but WITHOUT ANY WARRANTY.  No author or distributor accepts responsibility to anyone for the consequences of using them or for whether they serve any particular purpose or work at all, unless he  
says so in writing.  Everyone is granted permission to copy, modify and redistribute the programs in GSLIB, but only under the condition that this notice and the above copyright notice remain intact.