skip to navigation
skip to content

karta 0.9.0

Geospatial analysis in Python


Karta is a package for spatial analysis in Python. It simplifies geospatial data processing by providing efficient generic classes for vector and raster data sources, as well as a selection of analysis functions.

Vector data types

Data are represented as Point, Line, Polygon, Multipoint, Multiline, and Multipolygon instances.

All data contain a .crs member encoding coordinate reference information. All vector geometries possess a .properties dict containing free-form metadata. Multipart geometries additionally possess a .data member which is a simple typed table-like data structure.

Geometries implement methods for computing distances, directions, and spatial characteristics. Multipart geometries support fast spatial indexing through quadtrees and r-trees.

GeoJSON and ESRI shapefile formats are supported for reading and writing. Experimental support for GPX XML files is in the karta.vector.gpx submodule.

Vector geometries implement the Python `__geo_interface__ attribute <>`__ for vector geometries. This permits data to be exchanged between Karta and external modules that also implement __geo_interface__ (e.g. shapely, fastkml).

Raster data types

The primary raster data type is the RegularGrid, which represents one or more 2-d arrays of pixels spaced via an affine transformation. RegularGrids are backed by one of several Band implementations, with the default implementation using the blosc compression library for efficient in-memory storage. There is experimental support for disk-backed storage via GDAL.

Grids may be queried, resampled, sliced, masked, and merged. Arbitrary array-based functions may be mapped to raster data with RegularGrid.apply(). Raster functions including slope, gradient, and hillshade are in the karta.raster.misc submodule.

GeoTIFF images are the primary supported format, however ESRI ASCII grids may also be used (with limitations due to the format).

Coordinate reference systems

Data in Karta is referenced to positions on earth via CRS objects that implement projection and geodetic methods. Coordinate reference systems may be either geographical or projected.

Geographical CRS objects return spatial relationships in terms of the true computed distances and azimuths on a spherical or ellipsoidal Earth.

Projected CRS objects (e.g. UTM, Polar Stereographic, and Web Mercator) return spatial relationships in terms of a flat plane, dependent on the projection.


Read or create vector geometries:

point = Point((-130.0, 52.0), crs=LonLatWGS84)
line = read_geojson("linedata.json")
polygon = Polygon([(-515005.78, -1301130.53),
                   (-579174.89, -1282271.94),
                   (-542977.83, -1221147.82),
                   (-437864.05, -1251641.55),
                   (-438160.72, -1252421.48),
                   (-437961.28, -1285314.00)],

Perform simple queries:

point2 = Point((-25.0, 48.0), crs=LonLatWGS84)
point.distance(point2)          # Distance in geographical units
line.intersects(polygon)        # True or False
ch = polygon.convex_hull()      # Returns a new polygon

Load and manipulate raster data:

grid = read_gtiff("landsat_scene.tif")  # Leverages GDAL
grid.profile(line)              # Collect data along a line
grid.resample(500.0, 500.0)     # Return a grid resampled at a new resolution


Karta currently supports Python 2.7 and Python 3.4+.

The easiest way to install is via pip. Installation requires a recent version of setuptools.

pip install -U setuptools
pip install karta

Building from source

Building from source requires Cython and a C99-compliant compiler:

pip install Cython

Then, clone the repository and install:

git clone karta
cd karta/
python build


See the online manual, the tutorial, or read the API documentation.


Bug reports, feature requests, and pull requests are welcome.

Run unit tests with python tests/

The manual is built using Sphinx and requires numpydoc.

File Type Py Version Uploaded on Size
karta-0.9.0.tar.gz (md5) Source 2017-04-25 1MB