The reference layers must be loaded into ArcMap before starting this tool (see below).
The output of this tool is a 3D point shapefile that can be displayed in 3D in ArcScene.”
Defining the sampling volume
The x and y axis (2D) components of the sampling volume are simply defined by the extent of the reference polygon or raster layers. Furthermore, in the
case of raster reference layers, any NoData cell is assumed to be outside of the sampling space. Your reference layers will likely need to be
prepared for use in this tool by clipping them down to only the 2D extent you are interested in sampling.
The sampling space in the third dimension is defined in one of four ways:
– i) A constant z minimum and z maximum, applied to a polygon reference layer. Thus the sampling volume in this case is defined by the footprint of the
polygons in 2D, plus the minimum and maximum z bounds the user defines.
– ii) A variable z minimum and z maximum, applied to each polygon in the polygon reference layer. To use this option you must create two fields in the
polygon attribute table, one for the z minimum, and one for the z maximum. The fields can be populated so that each polygon has different z min/max
values. (Note that if all the polygons have the same z min and max values, then this operates in exactly the same way as option i) above). This option
clearly allows for the delineation of a much more complex 3D sampling space.
– iii) In reference to an elevation or bathymetry layer, represented by a raster surface. In this case, the 2D extent of the sampling space is the extent of
the reference raster, excluding any NoData cells that might be present within the bounds of the raster. The z axis bounds are delimited by the raster
surface and a constant z bound that you supply. For instance, in the case of bathymetry, we might wish to sample the volume of water between the surface
of the water (0m) and the sea floor. In this case bathymetry would be coded as negative numbers, and the user would supply zero as the other z bound on
– iv) The most complex but versatile method of defining the sampling volume is in reference to two rasters. In this case, the 2D footprint of the sampling
space is delimited by the overlapping area of the extents of the two rasters. (If one raster extends beyond the extent of the other, this tool will restrict
the sampling space to only the 2D area where the extents overlap. Again, NoData cells in either of the two rastes are excluded from the sampling space.
The bounds of the z axis is then defined by the two raster surfaces. The random points will be generated in between these two surfaces.
Note that in all of these cases, the sample produced is a pure random sample. At this time weighting and stratification are not possible, as they are
in the case of the 2D random point generation tool.