The problem
Most ways of describing a city assume that someone drew it first: streets with names, plots with edges, blocks that repeat. Informal settlements are not built that way, so the standard descriptors either fail outright or quietly measure the wrong thing. Anything you want to say afterwards about how air, light or heat move through that fabric depends on describing its geometry first.
What I build
A Python pipeline that turns building footprints and a digital terrain model into a grid of morphometric indicators, fine enough that a single alley is resolved rather than averaged away. The indicators are the standard vocabulary of urban physics, computed so they hold on a fabric they were not designed for: sky openness, frontal area density, height variation, slope, aspect, porosity.
On top of that sits a sampling design that decides where a computational fluid dynamics campaign should actually look, rather than simulating everywhere at uniform cost, and an ingestion layer that turns public meteorological records into the boundary conditions those simulations need. The simulations themselves run elsewhere; this work defines what they receive and how their output is read back.
What is shown here
The geometry of one site, Vidigal, is drawn on this site’s front page. Nothing else from this work appears here. It is unpublished research with co-authors, so this page describes the method and stops there: no indicators, no findings, no figures from the study.