topic Re: ArcGIS Raster in ArcGIS Spatial Analyst Questions

ArcGIS Raster

SamanArmal1 — Tue, 17 Nov 2015 18:03:10 GMT

Hi all,

I have boundary which has stored in a raster. The only available data in this raster, resides on he boundary itself, and there is no data available either inside or outside the boundary. I am trying to expand the raster beyond the boundary in a way that each row of cells has the same value the adjacent cell on boundary shows. Any idea how I can do this extrapolation?

Many thanks in advance,

Re: ArcGIS Raster

DanPatterson_Retired — Sat, 11 Dec 2021 13:23:36 GMT

Are you wanting to reflect the edge values like the following numpy example? and if so, how far inwards do you want to reflect/mirror?

>>> import numpy as np
>>> a = np.array([[1,1,1],[1,0,1],[1,1,1]])
>>> a
array([[1, 1, 1],
      [1, 0, 1],
      [1, 1, 1]])
>>> np.pad(a,1,mode="edge")
array([[1, 1, 1, 1, 1],
      [1, 1, 1, 1, 1],
      [1, 1, 0, 1, 1],
      [1, 1, 1, 1, 1],
      [1, 1, 1, 1, 1]])
>>>

Perhaps a better illustration

set all the cells within 2 cells of the boundary to 0
pad the raster by mirroring the edge by 2 cells

>>> c = np.arange(100,dtype='int32').reshape((10,10))
>>> c
array([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
       [20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
       [30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
       [40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
       [50, 51, 52, 53, 54, 55, 56, 57, 58, 59],
       [60, 61, 62, 63, 64, 65, 66, 67, 68, 69],
       [70, 71, 72, 73, 74, 75, 76, 77, 78, 79],
       [80, 81, 82, 83, 84, 85, 86, 87, 88, 89],
       [90, 91, 92, 93, 94, 95, 96, 97, 98, 99]])
>>> c[2:8,2:8] = 0
>>> c
array([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
       [20, 21,  0,  0,  0,  0,  0,  0, 28, 29],
       [30, 31,  0,  0,  0,  0,  0,  0, 38, 39],
       [40, 41,  0,  0,  0,  0,  0,  0, 48, 49],
       [50, 51,  0,  0,  0,  0,  0,  0, 58, 59],
       [60, 61,  0,  0,  0,  0,  0,  0, 68, 69],
       [70, 71,  0,  0,  0,  0,  0,  0, 78, 79],
       [80, 81, 82, 83, 84, 85, 86, 87, 88, 89],
       [90, 91, 92, 93, 94, 95, 96, 97, 98, 99]])
>>> d = np.pad(c,2,mode='edge')
>>> d
array([[ 0,  0,  0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  9,  9],
       [ 0,  0,  0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  9,  9],
       [ 0,  0,  0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  9,  9],
       [10, 10, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 19, 19],
       [20, 20, 20, 21,  0,  0,  0,  0,  0,  0, 28, 29, 29, 29],
       [30, 30, 30, 31,  0,  0,  0,  0,  0,  0, 38, 39, 39, 39],
       [40, 40, 40, 41,  0,  0,  0,  0,  0,  0, 48, 49, 49, 49],
       [50, 50, 50, 51,  0,  0,  0,  0,  0,  0, 58, 59, 59, 59],
       [60, 60, 60, 61,  0,  0,  0,  0,  0,  0, 68, 69, 69, 69],
       [70, 70, 70, 71,  0,  0,  0,  0,  0,  0, 78, 79, 79, 79],
       [80, 80, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 89, 89],
       [90, 90, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 99, 99],
       [90, 90, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 99, 99],
       [90, 90, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 99, 99]])

>>> import arcpy
>>> raster = arcpy.NumPyArrayToRaster(d)
>>> raster.save("f:/temp/raster.tif")

with some random color scheme

Re: ArcGIS Raster

SamanArmal1 — Tue, 17 Nov 2015 19:24:11 GMT

Thank you for your respond. Yes, its generally the same procedure. The only difference is that I am looking for producing cells with the same values.

Re: ArcGIS Raster

DanPatterson_Retired — Sat, 11 Dec 2021 13:23:39 GMT

You can pretty well do what you want, the documentation is good, here is a small sampling

>>> a
array([[1, 1, 1],
       [1, 0, 1],
       [1, 1, 1]])
>>> aa = np.pad(a,pad_width=2,mode='constant', constant_values=(3,4))
>>> aa
array([[3, 3, 3, 3, 3, 4, 4],
       [3, 3, 3, 3, 3, 4, 4],
       [3, 3, 1, 1, 1, 4, 4],
       [3, 3, 1, 0, 1, 4, 4],
       [3, 3, 1, 1, 1, 4, 4],
       [3, 3, 4, 4, 4, 4, 4],
       [3, 3, 4, 4, 4, 4, 4]])
>>> ab = np.pad(a,pad_width=2,mode='constant', constant_values=(5,5))
>>> ab
array([[5, 5, 5, 5, 5, 5, 5],
       [5, 5, 5, 5, 5, 5, 5],
       [5, 5, 1, 1, 1, 5, 5],
       [5, 5, 1, 0, 1, 5, 5],
       [5, 5, 1, 1, 1, 5, 5],
       [5, 5, 5, 5, 5, 5, 5],
       [5, 5, 5, 5, 5, 5, 5]])
>>>

Re: ArcGIS Raster

DarrenWiens2 — Tue, 17 Nov 2015 19:41:51 GMT

I believe the out-of-the-box ArcGIS tool for this is Euclidean Allocation.

Re: ArcGIS Raster

DanPatterson_Retired — Tue, 17 Nov 2015 19:55:34 GMT

Yes after you expand the extents, change nodata to some value then allocate, but It won't give you a constant value unless your edge values are constant initially. And SA doesn't handle the interior clipping as handlily as numpy. Some things, even if it is prep work, are better done outside SA.

Re: ArcGIS Raster

DarrenWiens2 — Tue, 17 Nov 2015 19:58:15 GMT

I interpret this as wanting to smear the boundary values, which Euclidean Allocation should do.

Re: ArcGIS Raster

DanPatterson_Retired — Tue, 17 Nov 2015 20:07:26 GMT

I read that he wanted a constant value around the edge...in any event, I can create any kind of value around the edge just as easily, you can even apply functions, like random etc

His post from above

"Thank you for your respond. Yes, its generally the same procedure. The only difference is that I am looking for producing cells with the same values."