Skip to content
This repository has been archived by the owner on May 8, 2021. It is now read-only.

Latest commit

 

History

History
191 lines (139 loc) · 7.34 KB

install.md

File metadata and controls

191 lines (139 loc) · 7.34 KB
Raw

Installation Instructions

If you are planning to participate in the "Geospatial data in Python: Database, desktop, and the web" tutorial associated with this Github repository, the following are the suggested installation instructions. We will take time during the session to go over installation issues, but if possible, at least try to have some of these installation steps completed before we start. If you are unsure of how/where to start, no problem, just show up to the session (or better yet, the 'install party' beforehand) and we'll get you going!

Base Install

  • First, check the Known Issues below for OS-specific Initial Setups.

  • Install Anaconda (I'm using Anaconda-2.1.0 [64-Bit - Python 2.7])

  • Create a new virtual environment (Skip this step on Windows, trust me, it'll be easier):

    ```bash
conda create -n pygeo pandas ipython-notebook matplotlib
source activate pygeo
```

  • Install pip for later (pip allows us to install additional Python packages not available via conda: bash conda install pip

  • If you don't already have it, you might need to install cython

    ```bash
conda install cython
```

  • Install required packages (on Windows, you might need to install binaries from here for shapely, pyproj, and rasterio).

    ```bash
pip install shapely
pip install fiona
pip install PIL
pip install pyproj
pip install descartes
pip install rasterio
```

    • You can use pillow in place of PIL if you like.

    • If any of the above commands cause an error, you can try using conda instead (replace PACKAGE below with the package you are trying to install):

      conda install PACKAGE

    • or check to see if a conda package exists using binstar:

      conda install binstar
binstar search -t conda PACKAGE

      • Make sure you find one for your OS. You can get more info about a package using the following command, which will also explain how to install the package:

        binstar show <USER/PACKAGE>

      • For example, you could used the following to install pyproj on OS X:

        conda install --channel https://conda.binstar.org/asmeurer pyproj

  • Install geopandas (important!)

    ```bash
pip install geopandas
```

(Web)mapping Packages

  • Install cartopy (on Windows, use binaries from here for cartopy.)

    pip install pyshp
pip install cartopy

  • Install mplleaflet (for making slippy maps). See the Known Issues below about install git.

    pip install git+git://github.com/mpld3/mplexporter.git
pip install git+git://github.com/jwass/mplleaflet.git

  • Install geojson.py for shooting data to the web!

    pip install git+git://github.com/jwass/geojsonio.py.git

Optional Packages

  • Install basemap, a common package for making static maps (I didn't install this):

    conda install basemap

  • Install psycopg2 for interacting with PostGIS (We don't need this, but I will do a demo with this):

    pip install psycopg2

Install QGIS

  • Go to the official QGIS page for details, or install via brew on OSX, apt-get on Linux, or OSGeo4W on Windows (Either way, this install will likely take quite a while).

    • I installed this on OSX via homebrew with:

      brew tap osgeo/osgeo4mac
brew install qgis-26

    • On Linux, if you follow the Linux Initial Setup below first, you should be able to install QGIS with:

      sudo apt-get install qgis

    • On Windows, follow the instructions on the official QGIS page.

Alternative Install Guide

Known Issues

  • In some cases, it may be better to pip install shapely than to conda install shapely, particularly when using cartopy.
  • If cartopy and shapely don't place nice together, it may be because of version issues. Please take a look here for a possible fix: #3

Linux Issues

  • Initial Setup

    sudo apt-get install git
sudo add-apt-repository -y ppa:ubuntugis/ppa
sudo apt-get update
sudo apt-get install -y gdal-bin libgdal-dev

    • On newer versions of Ubuntu (14.04 +), you may want to use the unstable repo (which is actually quite stable):

      sudo add-apt-repository ppa:ubuntugis/ubuntugis-unstable

    • On vanilla Ubuntu, you might need to install g++ before installing rasterio and others:

      sudo apt-get install g++

OS X Issues

  • Initial Setup

    • First install brew: ruby -e "$(curl -fsSL https://raw.github.com/Homebrew/homebrew/go/install)"

    • Then update and install gdal (You might be able to skip this step if you conda install fiona [see above]).

      brew doctor
brew update
brew install git
brew tap osgeo/osgeo4mac
brew install gdal
brew install qgis

  • On Mavericks+ if you don't already have developer tools installed, pip install pyproj will probably fail (due to missing gcc) and then ask you if you want to install them, so click 'yes' and then rerun pip install pyproj.

  • In some cases, importing shapely on OS X might fail while loading the GEOS library: bash OSError: Could not find library c or load any of its variants.

    • This can be fixed by using a newer version, or worked around by setting the following environment variable (add to your .bash_profile; see this issue for details):

      export DYLD_FALLBACK_LIBRARY_PATH=$(HOME)/lib:/usr/local/lib:/lib:/usr/lib

Windows Issues

  • Initial Setup

    • Download and install GDAL from here
    • If you use this one, then you should also use the fiona binary from that site. Otherwise, you'll have to play around with system PATHs etc. Others have had success with OSGeo4W, which includes many important libraries and their Python bindings.

  • If you don't already have it, you'll need to install git:

    • Download and install git from here.
    • When installing, make sure you choose to "Use Git from the Windows Command Prompt" (You may also want to install optional Unix tools).
    • You can also download GitHub for Windows here.

  • On Windows, source is not needed when activating a virtual environment if you are using the Anaconda Command Prompt: activate scipygis.