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

Re: GEOSTATS: Boundary delimiter algorithm!

Expand Messages
  • roswell marjorie
    Check out the bottom of http://www.sfiltd.com/products.htm The Convex Hull Generator creates the polygon which you would get if each datapoint were a nail in
    Message 1 of 4 , Feb 7, 1997
      Check out the bottom of

      http://www.sfiltd.com/products.htm

      "The Convex Hull Generator creates the polygon which you would get if each
      datapoint were a nail in a board and youstretched a rubber band around the
      outside of all of them. "

      This is a MapInfo-based solution.


      Margie


      On Fri, 7 Feb 1997, Paloma Urbano Lopez de Meneses wrote:

      > Hello to everyone!
      >
      >
      > I am trying to delimit a boundary out of a disperse set of points
      > considering only the most external ones.
      > The example should be something like the following,
      >
      > + +
      > +
      > + +
      > +
      > + +
      >
      > This are the disperse points
      >
      > The algorithm will delimited the outside boundary.
      >
      > +-----+
      > / + \
      > / + +
      > + |
      > \__ +____+
      >

      _________________________________________________________
      Marjorie Roswell, Spatial Analyst
      UMBC Center for Health Program Development and Management
      1000 Hilltop Circle Fx: (410)455-6850
      Baltimore, MD 21250 E: roswell@...
      Ph: (410)455-6802 http://umbc.edu/~roswell/mipage.html
      _________________________________________________________

      --
      *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.
    • Roberto Bruno
      ... Dear Paloma, I cannot help you about generic algorithms you ask for, but I suggest you to make sound use of: a) the phisical meaning of the boundary
      Message 2 of 4 , Feb 7, 1997
        paloma@... wrote:
        >
        > Hello to everyone!
        >
        > I am trying to delimit a boundary out of a disperse set of points
        > considering only the most external ones.
        > The example should be something like the following,
        >
        > + +
        > +
        > + +
        > +
        > + +
        >
        > This are the disperse points
        >
        > The algorithm will delimited the outside boundary.
        >
        > +-----+
        > / + \
        > / + +
        > + |
        > \__ +____+
        >
        > At the moment we have been triangulating the points and obtaining the
        > outline, however we are interested in an algorithm which would
        > recognize the most external points of the area and use them to
        > create a boundary polygon.
        >
        > Any help on this, useful web pages or literature would be appreciated.
        >
        > Thank you in advance,
        >
        > Paloma Urbano

        Dear Paloma,

        I cannot help you about generic algorithms you ask for, but I suggest
        you to make sound use of:
        a) the phisical meaning of the boundary polygon that you are looking for;
        b) the nature of data linked to the locations.
        In effect in many cases you could get good results by just exploiting the tools
        of geostatistics (of course used properly) and the selection criteria.
        For example, if in each point you have a thickness value of a geological body
        and you are looking for an evaluation of the total volume, you could make a
        kriging of the thickness at the nodes of a grid superimposed to the potential
        interest area.
        The boundary could then be defined on the base of the selection made on
        kriged values at the grid nodes. You just have to adopt one (or more) sound
        criteria :
        - a cut off value on the estimation variance (absolute or relative);
        - a selection of nodes with a consistent probability of having a positive
        thickness, but considering a lower limit of a confidence interval
        [Z*- k sigma2(Z*->Z), k>=0.];
        - the range value of variogram model of the variable at hand;
        - ... and so on, depending on the nature of your problem (a "noise level" is
        a variable quite different from "thickness" or from a "proportion", and so is
        the associated problem).

        Even if you have not the actual values of your (regionalized) variable,
        in my opinion you should refer to the true nature of your problem in choosing the
        boundarying procedure.

        I suggest you to address a specific/technical literature of your working field,
        and to examine not a "blind" or generic procedure, but that one which you
        select as the most proper when looking at the heart of your problem.

        Best regards.

        Prof. Roberto BRUNO
        Dip. Chemical, Mining & Environmental Technlogies Engineering
        Univ. of BOLOGNA, Italy
        Tel.(+39-51) 6443393; fax (+39-51) 6443392
        E-mail: SCIENMIN2@...
        --
        *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.
      Your message has been successfully submitted and would be delivered to recipients shortly.