Hi Lennert,

 

It is possible to work with Arc rasters as numpy arrays. MGET does this frequently. I recommend the GDAL package; see http://www.gdal.org. The code will ultimately look similar to this:

 

import gdal, gdalconst

dataset = gdal.Open(pathToRaster, gdalconst.GA_ReadOnly)

band = dataset.GetRasterBand(1)

data = "band.ReadAsArray()        " # data is a numpy array

noDataValue = band.GetNoDataValue()    # always returns a float, even for integer rasters

 

Alternatively, you could use MGET to do it:

 

from GeoEco.DataManagement.ArcGISRasters import ArcGISRaster

data, noDataValue = ArcGISRaster.ToNumpyArray(pathToRaster)

 

The MGET function wraps GDAL code similar to what I show above and performs some additional services:

 

·         It can read rasters from geodatabases (GDAL only reads from rasters on disk)

·         It returns a numpy array that uses the most compact data type to represent the data (ArcGIS and GDAL occasionally select sub-optimal data types)

·         It returns an int NoData value for integer rasters (GDAL always returns a float, even for integer rasters)

 

For more on MGET's Python functions, view the GeoEco Python Reference under the Marine Geospatial Ecology Tools folder in your Start menu. In that Reference, click GeoEco Module Index and then click GeoEco.DataManagment.ArcGISRasters, and drill in until you find the appropriate function.

 

The choice of whether to use GDAL or MGET is up to you. If you are writing a tool that you plan to redistribute to others, you may prefer to use GDAL because it is lighter weight than MGET and has an MIT license (less restrictive than MGET's GPL). If you prefer to keep your code compact and like the additional services provided by MGET, you might prefer it instead.

 

GDAL does not allow you to write rasters, at least not in the traditional ArcInfo Binary Grid format or to geodatabases. It can write rasters in many other formats, such as ArcInfo ASCII Grid, GeoTIF, Erdas IMAGINE (.img). To write ArcInfo Binary Grids or to geodatabases, you can use the MGET ArcGISRaster.FromNumpyArray function. See the documentation I referenced above for more info.

 

Others will probably be interested in this answer. Do you mind if I forward this to the mget-help list? (Note: This will cause the message to be archived on our server for others to view. It will include your email address. I can redact that if you like.)

 

Jason

 

 

Hi Jason,

I have a question about raster processing in ArcMap via Python scripting.
I like to work with some rasters as arrays. Is this possible? The purpose is to average, exclude and find minimum and maximum values from several rasters. I like to do this I don't know the conversion code from raster to array.
(I already installed the numpy and scipy tools to work with arrays.)

I also wonder if there exists a decent manual for studying Python Scripting with ArcGIS.

Thanks a lot!

Lennert