Quickstart

This walkthrough analyses the example map that ships with Landschaft. It needs nothing but a working installation. Run the code in a Jupyter notebook, or save it in a .py file and run it with python.

1. Load a map

import landschaft as ls

forest = ls.load_raster(ls.example_data("input"))
print(forest.shape, forest.dtype)   # (1000, 1000) uint8
print(forest.rio.crs)               # its coordinate reference system

ls.example_data("input") is the path of a GeoTIFF inside the package; load_raster reads any raster file the same way, e.g. ls.load_raster("my_forest.tif"). The result is an xarray.DataArray: the pixel values plus the georeferencing (coordinates, CRS, NoData), which every Landschaft function keeps.

The map is coded the way habitat analyses expect it:

value	meaning
0	missing data (outside the study area, clouds...)
1	background: anything that is not habitat
2	foreground: the habitat, here forest

Your own maps rarely come coded this way; core concepts shows how to build such a map from a land-cover or tree-cover raster with ls.reclassify.

2. Look at it

ls.plot(forest)

ls.plot draws the map on an interactive map over satellite imagery: inline in a notebook, in a window of its own from a script. The plotting page shows how it draws every kind of data and how to choose the colours.

3. Analyse the habitat pattern

Morphological Spatial Pattern Analysis (MSPA) gives every habitat pixel a role: core (interior habitat), edge (the outer boundary), perforation (the boundary of holes inside the habitat), bridge (connecting two cores), loop (connecting a core to itself), branch (a dead end) and islet (too small to have a core).

morph = ls.mspa(forest, edge_width=2)
ls.plot(morph)                       # GuidosToolbox's colours and class names

edge_width=2 makes edges two pixels wide; everything further inside a patch is core. On a 1,000 × 1,000 map this takes a fraction of a second.

4. Summarise

stats = ls.mspa_stats(morph)
print(stats)

Each row is a class, with its number of pixels, its share of the data area and its share of the habitat. The table is a pandas DataFrame: save it with ls.save(stats, "mspa_classes.csv") (or .xlsx) or keep working with it in pandas.

5. Save the map

ls.write_raster(morph, "forest_mspa.tif")

The GeoTIFF keeps the georeferencing and embeds the MSPA colour table, so it opens in QGIS or ArcGIS with the usual colours.

Where next

• Core concepts: how to prepare your own data.
• The tutorial of your topic, from habitat patterns to circuit connectivity.
• Configuration: threads and memory for large maps.