GIS Day and World Town Planning Day · Nov. 12, 2026

Storytelling with Maps

Stephen Stirling

Data Editor, The Philadelphia Inquirer

About me

Stephen Stirling

  • Current data editor at the Philadelphia Inquirer, where I manage data for the newsroom and edit a team of three reporters and developers.
  • Graduated from Rowan University, Class of 2006. Served as editor-in-chief of the Whit.
  • Became a data journalist in 2011. Learned about GIS through editors, colleagues, conferences and a network of data journalists around the country.
  • Geographers, planners and journalists often find themselves in the same datasets with similar goals. I hope we can learn from one another.

About me

I started my journey with paper maps

As a pizza deliveryman, this is how I got around.

A Hagstrom folded paper street map of Newark, New Jersey, opened flat. Streets are drawn in fine gray lines, highways in pink and red, parks in green, Newark Liberty airport in yellow, and ZIP codes printed in large red numerals across the city and its neighbors. The folded cover and the street index are laid alongside it.
A Hagstrom street map of Newark, New Jersey — streets, highways and ZIP codes, with a street index on the back.Credit: Hagstrom Map

Maps in news

Maps in print

An 1896 New York Times campaign map supplement, yellowed with age, showing states colored yellow for Republican and blue for Democratic, with stacked pyramid-shaped tables of vote counts sitting on top of each state.
The New York Times campaign map supplement, July 26, 1896 — state-by-state returns stacked in pyramids on top of the geography. Credit: Tim Wallace
A printed USA Today weather page opened flat, showing a national forecast map of the United States shaded by temperature band, with four smaller precipitation forecast maps below it.
USA Today’s use of graphics and maps helped push the journalism industry toward more visual design.

The digital transition

From print to digital static maps to live data

Then

Print

Maps and visuals increase in use over the 20th century in the print industry, providing key context and a sense of place. As print production improves, they get more complex and sometimes drive coverage.

Next

Digital static

Mapping becomes ubiquitous in the internet age as technology improves and learning curves come down. Tools allow GIS to become more accessible.

Now

Live data

Maps update themselves with new data and can better cater to readers’ needs. Anyone with some determination can make a professional-level map.

At the Inquirer

Maps give our readers a sense of place

Inquirer map: Where the virus has spiked in the Philadelphia area, coronavirus case rates by municipality and ZIP code.
Inquirer map: Philadelphia neighborhoods where property assessments increased the most between 2025 and 2027.
Inquirer map: Vote shifts for president by New Jersey municipality, 2020 to 2024, drawn as arrows.
Inquirer map: Four days of shootings near SEPTA buses, four numbered locations across Philadelphia.
Inquirer map: Measles cases by county in Pennsylvania in 2026, shaded by case count.
Inquirer map: Kindergartens most vulnerable to a measles outbreak, schools sized by enrollment and colored by vaccination rate.
Inquirer map: Ocean City's beach Christmas trees, nine trees located along the barrier island.
Inquirer map: Top wines by city and type in Pennsylvania, sized by units sold.

Maps Philadelphia Inquirer staff

My approach

Start with data, not a map

  • We make lots of different kinds of visualizations and do lots of different analysis at the Inquirer. We use maps when they fit.
  • Interview your data. Find out what you have, what you’re missing. Think about how someone might interpret it visually.
  • Sometimes geography is the story, sometimes it isn’t.
  • For us, a map is a late decision, not an early one.

Data <> GIS

It’s a two-way street

Sometimes geography is the story, but a map still isn’t the answer. This is GIS work too.

Screenshot of an Inquirer analysis panel headed Neighborhood Plow Coverage, showing that 94.0% of Philadelphia street segments — 27,131 of 28,856 — were plowed at least once, with per-neighborhood rates from West Philadelphia/Parkside at 80.7% up to North Philadelphia/East at 92.2%.
Plow coverage by neighborhood, built from street-segment data rather than from a map.

Putting people in places

Every ICE transfer from NJ and PA

Loading map…

Loading map…

Open all transfers ↗

Putting people in places

Every person in NJ, by race

Loading map…

Open map ↗

Geographic analysis

Philly shifts right

Loading chart…

Open map ↗

Geographic analysis

Philly’s phones won’t stop buzzing

Loading map…

Showing change

Wildfire smoke

Live data

Preparing for and Covering Elections

Loading chart…

Open map ↗

Live data

The Philadelphia Weather Machine

Loading…

Illustrating cultural disputes

The true borders of North, Central and South Jersey

Loading map…

Open map ↗

AI <> GIS

A new frontier, with new heights and risks

  • AI brings new opportunities for building on skills.
  • AI is NOT a replacement for foundational knowledge.
  • We use AI to boost our work with transparency and oversight.

Exploring Isometric Philly

Executing an idea for this talk

Building Sim Philly

A SimCity screenshot looking down on a city at an angle: a cluster of glass and brick skyscrapers along the top, stepping down through mid-rise blocks and apartment buildings to detached houses, strip malls and tree-lined avenues in the foreground.
A screengrab from the video game Sim City.Credit: Maxis

Exploring Isometric Philly

It’s been built, but can it be built with public data?

Isometric illustrated rendering of Lower Manhattan from above, showing every building in the financial district, the Brooklyn and Manhattan bridges, and the parks and piers along the waterfront.
Lower Manhattan, rendered building by building.Credit: Andy Coenen
Isometric illustrated rendering of central London, showing the Thames, Westminster, the London Eye, Trafalgar Square, St James's Park and Charing Cross station.
Central London, same approach.Credit: Boris Starkov

Exploring Isometric Philly

Gathering the data

  • Building footprints, tree inventory, roads and rails — things that gave us our initial layout.
  • Aerial imagery, LiDAR, building height data — things that give us shape and distinction.
  • Property data for information on facades, windows, storefronts, etc. Public photos of landmarks to shape individual details.

Exploring Isometric Philly

Building a base model

Untextured three-dimensional massing model of Center City Philadelphia seen from above at an angle: grey and tan building volumes, the Benjamin Franklin Parkway and Logan Square fountain at left, and the skyscraper cluster around Market Street.
  • Work in segments; LiDAR data is very large.
  • Extrude buildings and other surfaces through footprint data and LiDAR.
  • Add trees, roads and other infrastructure.
  • Still missing: Facade/window/siding data.

Exploring Isometric Philly

Train the visual model (Qwen)

Isometric illustrated rendering of Philadelphia rowhouse blocks: long runs of brick houses with flat roofs, corner stores, street trees and a small park, drawn at an angle from above.
  • My one public data hiccup: Google Gemini was used to build models to direct Qwen.
  • Why use Qwen? It is an open-source image model. Qwen is supplied model images and all remaining data; draws pixel style across city segment-by-segment.

Exploring Isometric Philly

Start with a sample, and build out

Isometric illustrated rendering of a wider slice of Philadelphia, with the expressway curving along the top, a green public square, dense blocks of rowhouses and institutional buildings, and piers along the Delaware River at upper right.
  • Split the city into 15+ segments; LiDAR data for the city is about 850GB.
  • Get city borders framework and place finished sections within.
  • Each section takes 1-3 hours to process; $5-20 of GPU time, using Modal.

Exploring Isometric Philly

Refine, retrain on tricky territory

Your model will make ridiculous errors. Refinement is important.

Isometric rendering of a large institutional building complex set in parkland beside a river, with curving roads, dense tree cover and a bridge crossing at the bottom of the frame.
Isometric rendering of an industrial waterfront: long warehouses, rows of parked trailers, finger piers reaching into the river and a single tall apartment block at the water's edge.

Exploring Isometric Philly

Draw SimPhilly (which will be called Pixel Philly for copyright reasons)

Loading map…

Open map ↗

Exploring Isometric Philly

Explore it yourself!

Questions

Questions?

← → to navigate · F for fullscreen