Interactive Lab

Why City Boundaries Lie

The lines on the map are political, not economic. A city's official border was drawn by administration and rarely redrawn since — so it swallows farmland it shouldn't, and cuts off dense neighbourhoods it should contain. When your unit of analysis is "a city," that mismatch quietly distorts every comparison you make. This lab shows the problem, how satellite data fixes it, and why India in particular undercounts how urban it really is.

The administrative city vs the economic city

Economists define a city by agglomeration — a contiguous patch of built-up density where people and firms benefit from being close together. Administrations define it by whatever line was last drawn in a gazette. The two rarely match. Toggle the map to see the gap.

View:

Built-up density (where people actually are) Administrative border Economic city boundary

Toggle between the views above. Notice how the rigid administrative rectangle both includes farmland in one corner and excludes a dense new suburb — while the economic boundary hugs the actual built-up area.

This isn't hypothetical

Delhi & its edge

Noida, Gurugram and Ghaziabad are functionally part of one economic city with Delhi — shared labour markets, commutes and firms — yet sit in different states and administrations. Study "Delhi" by its border and you miss a huge chunk of the actual city.

Dhaka's mismatch

Dhaka's official limits reach out over farmland while dense, unplanned settlements at the fringe fall outside them. The border counts fields as city and leaves people out.

Every country, differently

"City" means something different in each country's law, so cross-country comparisons of "urban" are often comparing incomparable units — one reason global urban research is so hard.

How India undercounts its own cities

India's Census uses a strict, decades-old rule to decide what counts as "urban". Many places that are city-like in every practical sense stay classified as rural — so the official urbanisation rate is almost certainly an undercount.

The Census definition of "urban". A place is urban if it is either (a) a statutory town — has a municipality, corporation or cantonment board — or (b) a census town, which must meet all three of:

a population of at least 5,000;
a density of at least 400 persons per km²; and
at least 75% of the male main workforce in non-agricultural work.

Why this misses real urbanisation. The rule is gendered (it looks only at male non-farm work), it requires a place to be re-examined and re-classified (which lags reality), and a settlement can be dense, market-connected and effectively urban while a slice of male workers still report agriculture. The result: a fast-growing category of "census towns" and peri-urban settlements that live like towns but are governed and counted as villages — with rural budgets, rural schemes and no urban planning.

Boundary logic	What it counts as a city	Blind spot
Statutory (administrative)	Places a government chose to give an urban local body.	Politics and lag decide the line, not density.
Census town (India)	Meets the 5,000 / 400-per-km² / 75%-male-non-farm rule.	Gendered, agriculture-anchored, re-classified slowly.
Economic / built-up	A contiguous patch of built-up density, from satellite data.	Needs calibration; a snapshot in time; some manual tuning.

Why it's not just pedantry. How you draw the line decides who gets urban infrastructure, which government (rural or urban) is responsible, how big "urban India" looks to a planner or investor, and whether a comparison of "cities" across states or countries means anything at all.

Redrawing the city from space

Instead of trusting an administrative line, you can build a boundary from what the land actually looks like. The approach behind Development Data Lab's Global Urban Boundaries combines three satellite-derived signals and asks: does this small grid cell look urban?

Population density

How many people live in the cell — from gridded datasets like WorldPop.

Built-up area

How much of the cell is buildings and pavement — from GHSL (Global Human Settlement Layer).

Night-time lights

How brightly the cell glows after dark, a proxy for activity — from VIIRS.

The 2-of-3 rule. A cell is flagged urban if it clears the threshold on at least two of the three signals. Requiring two guards against false positives — a single bright highway or one dense but rural cell won't tip it. Crucially, the thresholds are calibrated country by country, not fixed globally, so the same method works in Manhattan and in densely-populated rural Bangladesh (where a single global density cutoff wrongly labels farmland as city). Contiguous urban cells are then stitched into one city.

Try it: classify a grid cell

Drag each signal relative to its (illustrative) threshold and watch the 2-of-3 rule decide. This is a simplified teaching model, not the production algorithm.

Population density 50

Threshold: 55 · below

Built-up share 50

Threshold: 50 · at/above

Night-time lights 50

Threshold: 45 · at/above

Population density clears threshold

Built-up share clears threshold

Night lights clear threshold

Honest about the limits. The method isn't fully automatic — some cities need manual fine-tuning, and thresholds for some countries are still uncalibrated. It's a single point in time, so it's a snapshot, not a film of growth. But as its authors argue, a semi-automated, density-based boundary reflects the real city far better than a decades-old administrative line.

Why any of this matters for your work

Boundaries are not neutral bookkeeping. The line you use silently shapes what you measure, who is served, and whether your comparisons hold up.

The line changes the finding

Per-capita anything. Include farmland and the denominator swells, so income, density and service coverage per person all look lower than the real city's.
Comparisons across cities or countries. If "city" means different things in each place, a ranking of "most productive cities" may just be ranking how boundaries were drawn.
Who is responsible. A dense settlement classified rural gets rural governance and budgets — no urban planning, no municipal services — even as it fills with people.
Growth vs annexation. A city can "grow" on paper simply because its border was redrawn. A fixed-boundary, satellite snapshot lets you separate real densification from administrative expansion.

The practitioner's habit. Whenever a claim rests on "the city," ask: which boundary — administrative or economic — and who drew it? State it explicitly in your method. If the answer changes your number, that's not a footnote; it's a finding.

Get the real data

This lab teaches the idea. For the actual boundaries and the tools to work with them, go to the source — Development Data Lab, an independent research group whose datasets are open and free:

Global Urban Boundaries 2026

Economic city boundaries for 5,000+ cities worldwide, with an interactive explorer comparing them to older definitions, and downloadable data.

Open DDL Urban Boundaries ↗

SHRUG (for India)

DDL's open, village/town-level platform linking Census, night-lights and more — the finest-grained way to study urbanisation in India. Also in our Indian Data Navigator.

Open SHRUG ↗

The Global Urban Boundaries dataset, the interactive explorer, and SHRUG are the work of Development Data Lab, not ImpactMojo. This lab is an independent teaching companion; all credit for the data and method is theirs.

Now make the call for your own work

Turn the lesson into a decision you can defend. Record which boundary your analysis will use and why, then export a short boundary-decision note to paste into a concept note, proposal, or methods section.

Your project or analysis

The city / place in question

Which boundary will your analysis use?

Why — and what is your unit of analysis?

A blind spot you'll flag in your write-up

The lesson and making the call are free. Exporting the note is a Premium feature (Practitioner plan and up).

You'll never trust a city border blindly again

Administrative vs economic boundaries, India's urban undercount, and how satellites redraw the map — enough to read urban statistics critically and pick the right unit for your own analysis.

Urban Geography Agglomeration Census Towns Satellite Data Data Literacy

Indian Data Navigator → Explore more Labs