I'm having a large dataset of forest polygons and would like to scatter individual tree locations over these areas. Here is how I have done this so far:

enter image description here

In "Create Random Points" number of points has been set to 999999 and the minimum allowed distance (linear) to 5 meters.

Why am I running the "create random points" 4 times in that model? The problem is that when running it once in ArcGIS, the points created are way too sparse. Running it four times and then merging all positions that are closer together than 1.4 meters (which essentially comes out to ~5 meters as tests have shown) gives me a much denser point cloud.

Any ideas how to change this approach into something faster with improved performance?

  • Is this purely for visualisation purposes? Or do you need auto-generated point features to run some kind of hypothetical analysis? Commented Aug 29, 2018 at 5:02
  • It's for visualization (placement of 3D tree models)
    – Brief
    Commented Aug 29, 2018 at 5:38

1 Answer 1


You dont mention how fast your model executes. I tried code below for three polygons, generating about 60000 random trees in one minute. Make sure not to input impossible combination of tree density and minimum tree distance or it will run forever.

import arcpy
from random import randint
arcpy.env.overwriteOutput = True

#Change these five lines
arcpy.env.workspace = r'C:\Folder\Default.gdb'
stands = r'Stands'
output_points = r'Randomtrees' #Will be created in script
trees_per_areaunit = 0.1 #Desired tree density
min_treedistance = 2 #Min distance between trees

arcpy.CreateFeatureclass_management(out_path=arcpy.env.workspace, out_name=output_points, geometry_type='POINT', spatial_reference=stands)

def giverandomcoord(xmin,xmax,ymin,ymax):
    return (randint(xmin,xmax),randint(ymin,ymax))

def givenogoarea(x, y, sigma):
    neighborhood = []
    X = int(sigma)
    for i in range(-X, X + 1):
        Y = int(pow(sigma * sigma - i * i, 1/2))
        for j in range(-Y, Y + 1):
            neighborhood.append((x + i, y + j))
    return neighborhood

with arcpy.da.SearchCursor(stands,['SHAPE@','OID@']) as cursor:
    for row in cursor:
        coords = []
        nogo = set()        
        ext = row[0].extent
        treegoal = int(ext.polygon.area*trees_per_areaunit)
        xmin, xmax, ymin, ymax = int(ext.lowerLeft.X), int(ext.upperRight.X), int(ext.lowerLeft.Y), int(ext.upperRight.Y)
        while len(coords)<treegoal:
            point = giverandomcoord(xmin, xmax, ymin, ymax)
            if point not in nogo:
                nogo.update(givenogoarea(point[0], point[1], min_treedistance))
        pointlist = [arcpy.PointGeometry(arcpy.Point(*c)) for c in coords]        
        arcpy.CopyFeatures_management(in_features=pointlist, out_feature_class=r'in_memory\temppoints')
        arcpy.MakeFeatureLayer_management(in_features=r'in_memory\temppoints', out_layer='templyrpoints')
        arcpy.MakeFeatureLayer_management(in_features=stands, out_layer=r'templyrstand', 
        arcpy.SelectLayerByLocation_management(in_layer=r'templyrpoints', select_features=r'templyrstand')
        arcpy.Append_management(inputs=r'templyrpoints', target=output_points)

enter image description here

  • 1
    Thank you very much! No matter what I set the maximum number to, the Random Points tool doesn't create something that leaves some areas blank. When creating random points in Global Mapper for example the result is much denser with a better scattering over the area (but performance is horrible, what takes minuted in ArcGis, takes days in GM). I will look into your script once I got Python working here. Need to reinstall ArcGis after finishing a few processes and see if Python works again then. Maybe it makes a difference to using the Random Points tool within the GUI?
    – Brief
    Commented Aug 31, 2018 at 2:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.