topic Re: Has anyone successfully created a script tool with an R script using a raster dataset? in R-ArcGIS Questions

Has anyone successfully created a script tool with an R script using a raster dataset?

StacyStrickland — Thu, 29 Sep 2016 17:36:28 GMT

I have an R script that uses the spatialEco package to create percent volume rasters. It works well as a stand-alone script but I would like to create a script tool in ArcGIS to use in ModelBuilder. The error I am getting seems related to the fact that the 'raster.vol' function needs to operate on a raster layer object but when I use the 'arc.open' function in my script, my raster opens as a :RasterDataset and arcgisbinding doesn't seem to have an inherited method to allow me to use the 'raster' function to convert the object to a raster layer. Any suggestions?

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

SteveKopp — Thu, 20 Oct 2016 16:15:45 GMT

Stacy - raster data is not currently supported in the the R-ArcGIS Bridge. This is functionality we are adding so if you have particular requirements for raster support please let us know.

Steve

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

StacyStrickland — Thu, 20 Oct 2016 18:35:36 GMT

Thanks for the reply, Steve! Looking forward to when raster data is supported to be able to automate some processing.

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

DanPatterson_Retired — Thu, 20 Oct 2016 18:43:36 GMT

what are you looking to do since the numpy and scipy covers most of R... depending what you are working on

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

StacyStrickland — Thu, 20 Oct 2016 20:56:33 GMT

I would like to create a tool to use in Model Builder with other tools to automate a series of processes that involves creating a raster in ArcGIS then running a script in R and adding the results back into ArcGIS. See my original post for the exact function I am using in R.

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

DanPatterson_Retired — Thu, 20 Oct 2016 22:20:48 GMT

Yes I saw that it is product by some spatialEco and does % raster volume which is why I asked since it sounded pretty generic and something that could be accomplished with the aid of 3D or pure numpy.

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

StacyStrickland — Sat, 22 Oct 2016 16:13:43 GMT

Dan, I'm not familiar with 3D or numpy, so I don't know if they are capable of the same raster calculation. Something to look into, thanks!

Re: Has anyone successfully created a script tool with an R script using a raster dataset?

DanPatterson_Retired — Sun, 12 Dec 2021 09:25:54 GMT

There is some pretty simple stuff you can do with numpy. Consider a slope with elevations ranging from 0 to 29 m and a cell size of 1m^2. What volume of earth would be removed if we removed all the material >=15 leaving behind a flat plateau behind.

In [15]: a = np.arange(5*6).reshape(5,6)
Out[16]: a
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]])

In [17]: a[a>15]
Out[17]: array([16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29])

In [18]: np.sum(np.abs(15 - a[a>15]))
Out[18]: 105
‍‍‍‍‍‍‍‍‍‍‍‍‍

So if you have a specific example, I am sure it can be done. You aren't limited to 1d or 2d arrays either. For instance, consider a tiny place (2x2) over 5 time periods. How do you calculate the per area averages over the time period, the averages per time and the overall average.

In [19]: t = np.arange(5*2*2).reshape(5,2,2)

In [20]: t
Out[20]: 
array([[[ 0,  1],
        [ 2,  3]],

       [[ 4,  5],
        [ 6,  7]],

       [[ 8,  9],
        [10, 11]],

       [[12, 13],
        [14, 15]],

       [[16, 17],
        [18, 19]]])

In [21]: np.average(t, axis=0)
Out[21]: 
array([[ 8.000,  9.000],
       [ 10.000,  11.000]])

In [22]: np.average(t, axis=1)
Out[22]: 
array([[ 1.000,  2.000],
       [ 5.000,  6.000],
       [ 9.000,  10.000],
       [ 13.000,  14.000],
       [ 17.000,  18.000]])

In [23]: np.average(t)
Out[23]: 
9.5‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

R and Numpy do share a lot... just a bit of difference in syntax.. there are a few translation tables out there converting between R, Numpy and Julia for instance. I have a large repository of 'stuff' on my blog if you are interested.