Global Mapper v25.0

Gridding algorithm

miksmith
miksmith Global Mapper UserTrusted User
edited December 2007 in Technical Support
The gridding algorithm for point elevation data (in this case LiDAR) suggests that gridding algorithm generates a TIN which I assume is then converted to a raster. Whilst TINs are the best method to store and manipulate DEMs from original point data, they are not a particularly good method to generate output rasters. Surfer for example has options for kriging, splines and RBFs amongst the usual nearest neighbour etc.

Unless Ive missed it in the manual, I havent seen an explicit section detailing how the raster conversion works for this type of data.

ANy comments most welcome.

thanks

mike

Comments

  • global_mapper
    global_mapper Administrator
    edited December 2007
    Mike,

    Global Mapper only uses the TIN as an intermediate step to generate the raster. Once a TIN is generated internally, Global Mapper then determines an optimal grid size (if necessary), then steps through each point in the grid that will be generated and determines the elevation to assign that grid point based on the TIN. At the end of this operation the TIN is discarded unless you checked the option to keep it as a separate layer.

    Let me know if I can be of further assistance.

    Thanks,

    Mike
    Global Mapper Support
    support@globalmapper.com
  • miksmith
    miksmith Global Mapper User Trusted User
    edited December 2007
    Hi Mike

    That's most helpful. Two further queries:

    -how is the optimum cell size calculated?

    -given the above, is the elevation value assigned to the cell based upon the value of the TIN facet for the centre point of the cell?

    Any thoughts of have several different algorithms available? Whilst GM is clearly not meant to be competing with Surfer, splines make very pleasing looking rasters and, for many purposes, give sensible gridded values. Kriging is probably not a good idea given the relative complexity of the modelling process.

    Anyway, a thought! Regardless, we can always round trip such things through R.

    thanks

    mike
  • global_mapper
    global_mapper Administrator
    edited December 2007
    Provided that you have not manually provided a cell size to use, Global Mapper creates one by examining the triangles in the TIN and setting the cell width to 2.5 standard deviations below the mean triangle width and the cell height to 2.5 standard deviations below the mean triangle height. This means that the grid cells will be smaller than about 98% of the triangles, thus maintaining the full resolution provided by the input sampling.

    The elevation obtained for each grid cell is obtained by intersecting a line in the Z plane at the XY location for the cell through the plane formed by the triangle containing the XY location of the cell. Thus you get different elevation value for different locations in each triangle.

    Let me know if I can be of further assistance.

    Thanks,

    Mike
    Global Mapper Support
    support@globalmapper.com
  • miksmith
    miksmith Global Mapper User Trusted User
    edited December 2007
    Hi Mike

    Very many thanks for the detail. Just to confirm, where the triangular facet/plane intersects the x,y of the cell's z-plane, the elevation is calculated using a linear interpolation between the vertices? (if Ive got that right in my head!!!)

    And the selection of cell size seems most appropriate.

    cheers

    mike
  • global_mapper
    global_mapper Administrator
    edited December 2007
    Global Mapper just does an intersection of a 3D line and a 3D plane to get the intersecting 3D point, then uses the Z value of that intersection as the cell elevation. I guess the math for this basically works out to a linear interpolation using the 3 corner points of the enclosing triangle.

    Let me know if I can be of further assistance.

    Thanks,

    Mike
    Global Mapper Support
    support@globalmapper.com