bluegeo is a library developed to assist in geospatial development where core scientific python libraries such as numpy, scipy, scikit-image and scikit-learn can be implemented easily over raster and vector geospatial data.

For example:

import bluegeo as bg
import numpy as np

# Virtually read a raster file
r = bg.Raster('elevation.tif')

# Slice the data, returning a numpy array
# Note: Slice-based __getitem__ operations are supported, but fancy indexing is not
a = np.ma.masked_equal(r[:], r.nodata)

# Perform a numpy operation
print(a.mean())
784.91772

bluegeo requires python 3, and a system installation of GDAL and GRASS

Clone the bluegeo repo

git clone https://github.com/bluegeo/bluegeo.git

from the root: pip install .

Build the image (from the root dir)

docker build -t bluegeo .

Start a session in the container

docker run --rm -v /home/ubuntu/bluegeo/scratch:/scratch -it bluegeo /bin/bash

Note:

• To preserve the container, do not use the --rm flag.
• A scratch directory (absolute path) is mounted to share files - omit it if necessary.

The easiest way to use bluegeo is on Linux, using the privision.sh script to install the dependencies:

bluegeo/provision.sh

Re-initialize conda & activate the environment:

eval "$(~/miniconda/bin/conda shell.bash hook)"
conda activate bluegeo

Test the installation

python

>>> import bluegeo as bg
>>>