Momepy is built on top of geopandas GeoDataFrame objects and, for
network analysis, on networkx Graph.
For any kind of morphometric analysis, data needs to be provided as
GeoDataFrames. Results of morphometric analysis from momepy can
be generally returned as pandas Series to be added as a column of
existing GeoDataFrame. All the detailes and attributes of each class
are clearly described in the API.
Morphometric functions available in momepy could be divided into
four different groups based on their approach to data requirements and
Simple morphometric characters are using single GeoDataFrame as a
source of the data.
Relational characters are based on relations between two or more
GeoDataFrames. Typical example is AreaRatio, which requires
a) features to be covered (e.g. land unit) and b) features which are
covering them (e.g. buildings).
Network analysis (graph module) characters are based on
networkx.Graph and returns networkx.Graph with additional
node or edge attributes.
Additional modules (elements and utils) cover functions
generating new morphological elements (like morphological tessellation)
or links between them. For details, please refer to the API.
Majority of functions used within momepy is not limited to one type
of morphological elements. However, the whole package is built with a
specific set of elements in mind, based on the research done at the
University of Strathclyde by the Urban Design Studies Unit. This is
true especially for morphological tessellation, partitioning of space
based on building footprints. Morphological tessellation can substitute
plots for certain types of analysis and provide additional information,
like the adjacency, for the other. More information on tessellation is
in dedicated section of this guide.
Generally, we can work with any kind of morphological element which fits
selected function, there is no restriction. Sometimes, where
documentation refers to buildings, other elements like blocks can be
used as well as long as the principle remains the same.
For example, you can use momepy to do morphometric analysis of:
When using more than one morphological element, momepy needs to
understand what is the relationship between them. For this, it relies on
unique_id attributes. It is expected, that every building lies on
certain plot or morphological cell, on certain street or within certain
block. To use momepy, each feature of each layer needs its own
unique_id. Moreover, each feature also needs to bear unique_id
of related elements. Consider following sample rows of
Each building has its own unique building_id, while more building
share block_id of block they belong to. In this sense, in
blocks_gdf each feature would have its own unique block_id used
as a reference for buildings_gdf. In principle, elements on the
smaller scale contains IDs of elements on the larger - blocks will not
have building IDs.
Momepy can generate unique ID using momepy.unique_id() and link
certain types of elements together
Unique IDs are also used as an ID within spatial weights matrices.
Thanks to this, spatial weights generated on morphological tessellation
(like Queen contiguity) can be directly used on buildings and vice
versa. Detailed information on using spatial weights within momepy will
be discussed later.