Raster representation in arcpy/numpy

Question

I have a .tif raster file which contains integer values at 1, 2, 3, 4 and 5. Importing this file into ArcMap also shows a number of cells with Nodata value where the above classes are not present.

I was able to convert a tile from this .tif file into a numpy array using the Arcpy documentation. I basically ran the following code

point = arcpy.Point(XCoord, YCoord)
test = arcpy.RasterToNumpyArray(RasterFileName, point, 10,10)

Where XCoord and YCoord are specifically selected to be in an area where all the returned raster values should be Nodata (according to ArcMap). Instead, I get the following output when I print out test:

array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]])

Why are these not the values corresponding to ArcMap?. For another set of coordinates in the region where there is data I get the following:

array([[2, 3, 2, 1, 2, 2, 2, 1, 1, 2],
 [3, 3, 2, 1, 2, 2, 2, 1, 2, 2],
 [3, 2, 2, 2, 2, 2, 2, 2, 3, 4],
 [3, 2, 2, 3, 2, 3, 3, 3, 3, 4],
 [2, 2, 3, 3, 2, 2, 3, 2, 2, 2],
 [1, 3, 3, 3, 2, 2, 2, 1, 1, 2],
 [1, 2, 2, 2, 2, 2, 2, 2, 1, 2],
 [1, 2, 2, 2, 2, 2, 3, 3, 2, 2],
 [1, 1, 1, 1, 1, 2, 4, 4, 2, 2],
 [1, 1, 1, 1, 1, 2, 3, 3, 3, 3]])

This corresponds to the ArcMap information in that tile. Therefore I am sure that the raster is converting correctly.

How does ArcMap differentiate between NoDATA and 0 valued raster cells when in numpy NoDATA values are not allowed? Is the returned numpy array the actual data or is it simply a higher level representation of something else?

• Bera: See: desktop.arcgis.com/en/arcmap/10.3/manage-data/raster-and-images/…
• Jon: Are you sure you're sampling your arrays from the same location? Maybe you've flipped your x,y coordinates, maybe there's a CRS mismatch, etc. I can say that the numpy array is the "actual data" rather than some kind of representation.

Answer 1

Below is some raster data (happens to be a flow direction grid) where white is NODATA. The Black point is the lower left corner and the black square is a 10 x 10 pixel extent such as yours

Sample data

If I run the following code:

point = arcpy.Point(441179,90512)
test = arcpy.RasterToNumPyArray("flwdir",point,10,10)

Then test is:

array([[ 2, 2, 2, 4, 0, 0, 0, 0, 0, 0],
 [ 2, 1, 1, 1, 0, 0, 0, 0, 0, 0],
 [ 1, 1, 1, 1, 0, 0, 0, 0, 0, 0],
 [ 1, 4, 1, 2, 0, 32, 128, 0, 0, 0],
 [ 1, 2, 0, 0, 0, 16, 1, 0, 0, 0],
 [ 1, 4, 1, 0, 0, 16, 1, 0, 0, 0],
 [ 1, 4, 2, 0, 16, 32, 1, 0, 0, 1],
 [ 4, 0, 0, 0, 4, 2, 4, 0, 2, 4],
 [ 1, 0, 0, 2, 2, 2, 4, 0, 0, 16],
 [ 2, 0, 2, 2, 2, 4, 8, 0, 0, 4]], dtype=uint8)

All the NODATA have been converted to zero.

If you read the help file for conversion function there is one last parameter you had not included nodata_to_value.

So if you run this code:

test = arcpy.RasterToNumPyArray("flwdir",point,10,10,-9)

the output is now:

array([[ 2, 2, 2, 4, -9, -9, -9, -9, -9, -9],
 [ 2, 1, 1, 1, -9, -9, -9, -9, -9, -9],
 [ 1, 1, 1, 1, -9, -9, -9, -9, -9, -9],
 [ 1, 4, 1, 2, -9, 32, 128, -9, -9, -9],
 [ 1, 2, -9, -9, -9, 16, 1, -9, -9, -9],
 [ 1, 4, 1, -9, -9, 16, 1, -9, -9, -9],
 [ 1, 4, 2, -9, 16, 32, 1, -9, -9, 1],
 [ 4, -9, -9, -9, 4, 2, 4, -9, 2, 4],
 [ 1, -9, -9, 2, 2, 2, 4, -9, -9, 16],
 [ 2, -9, 2, 2, 2, 4, 8, -9, -9, 4]], dtype=int16)

So by declaring what NODATA should encode into you can distinguish between real zero values.