contouring algorithm

From: Dennis Shea (shea AT XXXXXX)
Date: Wed Aug 28 2002 - 15:51:19 MDT

  • Next message: Keith Oleson: "segmentation fault"

    The default line contouring algorithm should *not* create contour
    lines that cross. Having contour spikes/cross-overs would be unexpected.

    It is possible that if the 2D smoother is activated
    (eg: res@cnSmoothingOn=True) then under unique circumstances
    contour lines will cross, in particular, if the tension
    resource (cnSmoothingTensionF) is set to some near zero number.
    (see: http://ngwww.ucar.edu/ngdoc/ng/ref/hlu/obj/ContourPlot.res.html)

    re: "another contouring algorithm"

    I'm not sure about another algorithm. Actually, if needed,
    I pass a light 2D smoother over the data. This means
    any smoothing is under my control.

       plot = gsn_...( wks, smth9_Wrap(data, 0.50, -0.25, True/False), res)

    where the smth9_Wrap is a function in contributed that 'wraps'
    the built-in smth9 function. The advantage of using smth9_Wrap
    is that it returns all the meta data.

    Either in default mode (no smoothing) or with smoothing
    (res@cnSmoothingOn = True( you should not have to change your contour
    levels or intervals. It only smooths the line (I think :-)).

    re: GrADS and Cressman scheme

    If Grads uses a Cressman scheme then this would be an implicit
    smoother, would it not? The smoothing would be some function
    of the scan radius. The larger the scan radius, the more the
    smoothing.

    Regards

    _______________________________________________
    ncl-talk mailing list
    ncl-talk@ucar.edu
    http://mailman.ucar.edu/mailman/listinfo/ncl-talk



    This archive was generated by hypermail 2b29 : Wed Aug 28 2002 - 15:56:42 MDT