Loading ...
Sorry, an error occurred while loading the content.

GEOSTATS: simulation software

Expand Messages
  • brgray
    Here s a slightly organized digest of the responses to my simulate query. I had forgotten to add the nonnormal --but received some followup comments re
    Message 1 of 1 , Feb 18, 2000
    • 0 Attachment
      Here's a slightly organized digest of the responses to my simulate
      query. I had forgotten to add the "nonnormal"--but received some
      followup comments re simulating nonnormal, spatially-correlated data.
      Thanks to all who responded

      I'll add that SAS's PROC SIM2D can apparently generate simulated normal
      data that is correlated in 2D; I haven't tested this procedure
      (discovered today...)


      FROM EDZER PEBESMA:

      gstat, http://www.geog.uu.nl/gstat/


      FROM JOSE QUINTIN CUADOR GIL:

      For simulating spatially correlated data you can use SgSim.exe from
      GsLib: Geostatistical
      Library.
      Good Luck

      José Quintín Cuador Gil
      Departamento de Computación
      Universidad de Pinar del Río, Cuba.
      <cuador@...>
      <cuador@...>


      FROM YETTA JAGER:

      Gstat is a gnu-software, freely distributed, for Unix or windows
      platforms.
      If you search on Gstat you'll find its home page. It does kriging or
      un/conditional simulation.


      NONNORMAL FOLLOWUP FROM YETTA:

      We are working with categorical data using indicator kriging. This
      would
      be a non-parametric option for you, but the consensus is that gaussian
      simulation
      with a normal-score transformation of the data works best (see Cressie
      book, among
      others). Edzar Pebesma, the developer of Gstat, recently added this
      transformation,
      but leaves it up to you to do the back-transformation, which is the more

      controversial part. I haven't tried it.


      FROM SEBAASTIAN LUYSSAERT:

      You can use GSLIB simulated annealing, The program starts with a seed
      image with a user specified histogram and 'improves' the image
      according to a user specified variogram by swapping data pairs.
      At the end you have a simulated spatial correlated image.

      For the application
      Deutsch, Journel,1998. GSLIB Geostatistical Software Library and
      User's Guide. Applied Geostatistics series. Oxford University Press.

      For more theory
      Goovaerts, 1997. Geostatistics for natural resources Evaluation.
      Oxford University Press

      For software
      http://www.gslib.com/

      Sebastiaan Luyssaert
      Laboratorium voor Bosbouw
      Universiteit Gent
      Geraardsbergse steenweg 267
      B-9090 Gontrode Melle


      NONNORMAL FOLLOW-UP FROM SEBASTIAN:

      If you mean with non normal that the data do not follow a Gaussian
      distribution the proposed software should work. The seed image built
      based on the user specified histogram can be built with any histogram
      because the histogram is red from a data file instead of being
      specified by its parameters.
      Sebastiaan Luyssaert
      Laboratorium voor Bosbouw
      Universiteit Gent
      Geraardsbergse steenweg 267
      B-9090 Gontrode Melle


      FROM MARC ROBSON:

      You may wan to have a look at MATLAB (www.mathworks.com) for your needs.
      It
      contains many tools useful for spatial simulation purposes, both raster
      and
      vector based. There are also a great many user written routines and
      toolboxes (groups of routines) which are available from the internet. My

      background is in remote sensing/GIS, I am utilizing Matlab to process
      imagery and simulate methods of ground cover estimation that are
      commonly
      used by ecologists and remote sensing researchers. I could use generic
      software for this purpose but MATLAB provides a far greater amount of
      flexibility for innovation.

      Aletrnatively if you run the Linux operating system you can have a look
      at
      the Matlab clone OCTAVE (www.che.wisc.edu/octave/octave.html)


      NONNORMAL FOLLOW-UP FROM MARC:

      Basically yes, many distributions other than normal are supported. From
      what
      I know about MAPLE, I don't think it would be as suitable. Very few of
      MATLAB's statistical and spatial tools would be covered. I think that
      MATLAB
      actually evolved from MAPLE, adding greater functionality and ease of
      use
      along the way so as to appeal to a larger market. As such, MATLAB has
      been
      adopted by a large section of the scientific and commercial community
      whereas, MAPLE has remained in the domain of mathematics and
      engineering.

      I should have mentioned that MATLAB also contains a sub-program called
      SimuLink, which is purely for running simulations. There is a student
      versionof MATLAB available that is cheap and contains most of the
      functionality. I believe the main limitation is the size of the arrays
      you
      can work with. This maybe a good way for you to examine the programming
      environment cheaply, if you are interested.


      FROM JEFFREY YARUS:

      There is quite a bit out there. If you are looking for free stuff (or,
      close to free), you might try GSLIB available on CD in a book published

      by Deutsch and Journel (Oxford University Press, GSLIS, Geostatistical
      Software Library). Lots of routines all in Fortran. Or, ISATIS by
      Geovariances (www.geovariances.fr). This will cost you some $$, but it

      looks like you are at a University so the it's very affordable. If
      you let me know more about your application, I might be able to offer
      you more advice. The two I mention are generic in application.

      For ISATIS, contact Brigitte Hogan in France (see website for tel.
      Number or email). For GSLIB, go to you college bookstore and order it.


      ****************************************************************
      * Brian R. Gray
      * Department of Epidemiology and Biostatistics
      * School of Public Health
      * University of South Carolina
      * Columbia, SC 29208
      * phone (803) 777-1765; fax (803) 777-8769; email brgray@...
      ****************************************************************


      --
      *To post a message to the list, send it to ai-geostats@....
      *As a general service to list users, please remember to post a summary
      of any useful responses to your questions.
      *To unsubscribe, send email to majordomo@... with no subject and
      "unsubscribe ai-geostats" in the message body.
      DO NOT SEND Subscribe/Unsubscribe requests to the list!
    Your message has been successfully submitted and would be delivered to recipients shortly.