Object-oriented Network Analysis Tools

РИДМИ (Russian)

ObjectNat is an open-source Python library for spatial analysis of urban objects (buildings, services, street observers, noise sources), developed by the IDU team at ITMO University. It computes accessibility, coverage and service provision on street graphs from IduEdu, visibility in 2D and over building heights, noise levels and insolation. Inputs and outputs are GeoPandas GeoDataFrames; the computationally intensive parts run in a compiled Rust extension.

Documentation · Examples · Документация на русском

Methods

Isochrones (docs, example). get_graph_isochrones and get_stepped_graph_isochrones return the area reachable from origins within a travel time (minutes) or network distance (metres) on an iduedu.UrbanGraph, as one polygon per origin or as bands of width step. The polygon is built from residual-distance buffers around the reachable nodes, at most off_network_distance (150 m) off the streets ("radius"), from the buffered reachable streets ("ways") or from merged buffers ("separate").

Coverage zones (docs, example). get_graph_coverage and get_stepped_graph_coverage assign every place to the service it reaches at the lowest network cost within a budget (search on the reversed graph, Voronoi cells of graph nodes), so zones do not overlap. get_radius_coverage is a graph-free variant with a fixed radius.

Service provision (docs, example). get_service_provision allocates building demand to services with limited capacity using a gravity-based probabilistic model over a dense or pandas-sparse cost matrix and a normative threshold. It returns a ProvisionResult with the sparse flow matrix and per-building and per-service metrics (provision_value is the share of demand supplied within the threshold); helpers join the metrics to GeoDataFrames, build link lines, drop links beyond a shorter threshold and clip the result to an area.

Visibility (docs, examples: 2D, 2.5D). get_visibility computes the exact visibility polygon (isovist) of each observer among obstacles treated as walls of infinite height and, with isovist_metrics=True, area, compactness, occlusivity, radial statistics, drift and visible share; get_isovist_grid maps these metrics over an area. get_viewshed uses building heights and terrain: a grid cell is visible when the line of sight from the eye (by default 1.6 m above the ground or a roof) to the cell passes above every wall and the ground. get_visibility_catchment and get_viewshed_catchment count, for every cell, the observers that see it (get_observer_points places observers along streets); rasterize draws the counts as a raster.

Noise (docs, example). simulate_noise_grid computes sound levels on a receiver grid from point, line and polygon sources: direct paths and first-order reflections with geometric spreading, air absorption, wall absorption and vegetation attenuation, and diffraction around building corners into their shadows, summed as energies. simulate_noise is a wave-front model for point sources with multiple reflections, and calculate_simplified_noise_frame a fast estimate from visibility masks and distance decay. Ground effect is not modelled. Background: wiki.

Insolation (docs, example). get_insolation follows the sun (NOAA solar position) through a day or a period and returns, on a ground grid among buildings of known height and optional terrain (GeoDataFrame, GeoTIFF or array), the daily hours of direct sun, including ground with no direct sun, the hours on the worst day and the longest continuous sunlit period, for which insolation norms set a minimum; rasterize draws them as a raster. get_facade_insolation computes the same quantities for facade points on every floor and checks them against a norm.

Rasters. Grid results (insolation and noise receivers, catchment cells, isovist fields from get_isovist_grid) are converted to rasters with rasterize; a Raster can be drawn with matplotlib, written to a GeoTIFF file or shown on a folium web map with explore.

Installation

pip install objectnat

ObjectNat supports Python 3.11–3.14 on Linux, macOS and Windows and installs IduEdu 2.4.2 or newer within the 2.x series. Wheels include the compiled extension objectnat._rust, so pip needs no compiler. Installing from a source checkout (pip install .) builds the extension and requires a Rust toolchain (rustup). GeoTIFF terrain requires the optional rasterio package.

Street graphs from IduEdu

Graph-based methods take an iduedu.UrbanGraph (nodes and edges as GeoDataFrames, edge weights time_min and length_meter) and use IduEdu’s Dijkstra routines; ObjectNat does not build graphs itself. IduEdu can download a Geofabrik extract and cut the territory out of it, so no Overpass request is needed (cutting requires pip install "iduedu[pbf]"):

import iduedu
from iduedu import get_4326_boundary, get_intermodal_graph

with iduedu.use_pbf(osm_id=1114252, region="northwestern-fed-district"):
    boundary = get_4326_boundary(osm_id=1114252)
    graph = get_intermodal_graph(territory=boundary, clip_by_territory=True)

Without the with block the same calls query the Overpass API.

Configuration

from objectnat import config

config.change_logger_lvl("INFO")   # hide debug messages
config.set_enable_tqdm(False)      # disable progress bars

Contacts

• NCCR, National Center for Cognitive Research

• IDU, Institute of Design and Urban Studies

• Natalya Chichkova, project manager

• Danila Oleynikov (Donny), lead software engineer

Contributions are described in CONTRIBUTING.md.

Citing ObjectNat

Citation metadata is kept in CITATION.cff; on GitHub, Cite this repository in the repository sidebar exports it as APA or BibTeX. Publications describing ObjectNat will be listed here.