The Technology

One map, many layers of meaning. This page explains, in plain language, what our geolocation mapping overlay platform does, why it is useful, and how it turns thousands of scattered data points into decisions you can explain, defend and repeat.

The Short Version

What this technology actually is

It is a map that knows things

Every clean-cooking installation sits on an interactive map at its real-world GPS position, accurate to about five metres. Around each pin, the platform stacks context: population density, official administrative borders, neighbouring countries and settlement type. It runs entirely in your browser on Leaflet, the open-source mapping engine behind some of the world's best-known map apps.

Information arrives in layers you control

Nothing on the map is decoration. Each layer answers one question well: where are people, where are stoves, where does this district end, did that unit move town. You switch layers on and off like channels, so the screen always shows exactly the amount of information the moment needs.

Every location carries a verified label

Urban, Peri-Urban or Rural is never guessed. Each record is classified against the United Nations' DEGURBA standard using a strict cascade of scientific datasets, and the winning source, threshold and date are written into the record itself. Challenge any label and the reasoning is right there.

Inference becomes almost visual

Because facts share one canvas, patterns surface without effort: dense neighbourhoods with thin stove coverage, quiet units clustering in one district, coverage stopping neatly at a national border. You spend attention on conclusions instead of assembling spreadsheets.

"A spreadsheet tells you what happened. The map shows you where, why it matters there, and what to do next."

How It Works

Five steps from GPS point to confident call

  1. Pin

    Each stove reports its position from the field. Coordinates land at roughly five-metre precision, tight enough to tell one side of a village from the other.

  2. Classify

    A build-time pipeline asks three scientific datasets in priority order: GHSL SMOD first, Kontur Population second, WorldPop third. The first source that can answer wins, and the decision plus its reasoning are written into the record.

  3. Layer

    The browser map stacks colour-coded markers, population density shading, clickable administrative boundaries and neighbour outlines on top of four switchable basemaps.

  4. Explore

    Click a region to filter everything to it. Toggle districts, density or neighbours as needed. Search, sort and drill into any pin; the dashboard's metrics recompute live as you go.

  5. Decide

    Target follow-up visits, set prices by market type, aim subsidies at underserved density, brief funders with evidence instead of assertions.

Layer by Layer

The overlay stack

Six layers work together. Each one exists to answer a specific kind of question, and each one is under your control.

Four basemaps

Street, Satellite, Dark and Terrain views of the same truth. Satellite shows ground reality, Dark makes coloured markers pop for briefings, Terrain explains why a road skips that village. Pick the backdrop that fits the question.

Installation markers with clustering

Hundreds of pins collapse into tidy clusters when zoomed out and expand smoothly as you zoom in. Colours carry meaning: blue for Urban, yellow for Peri-Urban, red for Rural. Coverage and gaps are visible before you read a single number.

Population density heat overlay

Hexagonal population cells shade from pale gold to deep red above 1,000 people per cell. Where markers are sparse but colour is hot, people outnumber stoves: that is where the next programme belongs.

Administrative boundaries

Official region and district borders from OCHA's Common Operational Datasets. Regions are clickable: selecting one filters markers, charts and metrics everywhere in the app. Districts toggle on when you need street-level accountability.

Neighbouring countries overlay

Dashed outlines of surrounding countries, each with its own checkbox. They stop border-region double-counting instantly and give cross-border programmes honest regional context.

Displacement lines

When a record has a second known location, a dashed line connects primary to secondary with a ghost marker at the far end. Stove moved to another town? You see it immediately instead of finding out at audit time.

Layer The question it answers The control it hands you
Basemaps ×4 What does this place physically look like? Switch backdrop without changing the story
Markers & clusters Where are the stoves, and what type of settlement? Filter by region, district, class, model or search text
Density heat Where are the people? Toggle on only when comparing need against coverage
Admin boundaries Whose jurisdiction is this? One click scopes the entire app to that region
Neighbours Is that really our deployment, or across the border? Per-country checkboxes, off by default
Displacement lines Did this stove move? Visual audit trail on the map itself

The Core Idea

Why overlays give definition and control over information

Flat data forces you to hold context in your head. A row reading "Fatick, rural" defines little and controls less. Give that same fact a position, a boundary, a density value and a colour, and it becomes defined: bounded, sourced, comparable. Give each of those meanings its own switch, and it becomes controlled: present exactly when you want it, gone when you don't.

🎯 Definition: every fact finds its place

A stove is no longer "somewhere near Fatick". It sits inside a verified border, within a population hexagon holding a known number of people, carrying a settlement class decided by a named dataset against a published threshold. Ambiguity has nowhere to hide.

🎚️ Control: you choose what's on screen

Every layer toggles independently. Click a region to scope everything at once. Tune classification thresholds from Settings without touching code, and the app prints the exact command to regenerate. The information load is your call, not the software's.

🧩 Context: inference gets easy

Decisions come from comparing things, and overlays put the comparison on one canvas. Stove clusters sit directly on top of the population shading they should match. Gaps, overlaps and outliers announce themselves visually before any chart confirms them.

🧾 Confidence: the trail is public

Labels ship with reasoning trails. Source datasets are version-pinned and checked live against provider APIs for updates. The full methodology document is published on purpose: if a funder asks how a village got labelled Rural, the answer is a link, not a shrug.

Before: flat spreadsheet era After: layered overlay map
"Somewhere in Fatick" → A five-metre GPS pin inside the officially verified Fatick border
"Rural", typed by whoever filled the form → A UN-standard class with the deciding source, threshold and date recorded
Context living in the analyst's head → Density, terrain and neighbours rendered on the same canvas as the data
One report per question, rebuilt each time → Filters and clicks reshape the same live registry in seconds

Who Uses It

From data to decisions, for four audiences

📋 Programme managers

"Where do the next 500 stoves go?" Compare hot density cells against marker coverage, filter to a region, and the KPI wall recalculates instantly. Market penetration and carbon density are computed for you.

🚐 Field teams

"Which villages do we visit first?" Inactive stoves rank themselves by region and district, and the registry travels offline: export to JSON on the last connected laptop, import when the network returns.

🔬 Analysts

"Can I trust these labels?" Yes, and you can attack them: thresholds are exposed in Settings, sources are pinned to versions, and every classification keeps its reasoning trail on the record for inspection.

💰 Funders & verifiers

"Prove the impact happened here." GPS-located demonstration placements, UN-aligned classification and published methodology turn impact reports from claims into evidence a verifier can walk through pin by pin.

Trust Built In

How every location earns its label

Classification follows the UN's DEGURBA standard (Statistical Commission, March 2020) through a strict priority cascade. The first source able to answer wins; the others stand by as fallbacks.

Priority Source & DEGURBA Term Threshold Applied
1 · Primary GHSL SMOD (EU Joint Research Centre), Urban Centre ≥ 1,500 people/km², cluster ≥ 50,000
2 · Fallback Kontur Population hexagons, Towns & Semi-dense Areas ≥ 300 people/km², cluster ≥ 5,000
3 · Validation WorldPop 100 m grid, Rural Grid Cells < 300 people/km²

Full technical detail, including SMOD code mappings and the compliance statement, lives in the methodology; provenance for every dataset is on Data Sources.

Plain Words

A friendly glossary

Term What it means here
Overlay A see-through layer of meaning stacked on the base map: markers, density shading, boundaries. Each one switches on and off.
Basemap The backdrop image (street, satellite, dark, terrain) beneath every overlay.
Cluster A bundle of nearby markers shown as one circle with a count until you zoom in.
Heat overlay Colour shading that shows where people live, cell by cell, so need and coverage can be compared by eye.
DEGURBA The UN's Degree of Urbanisation: the agreed statistical method for calling somewhere Urban, Peri-Urban or Rural.
GHSL SMOD The EU's Global Human Settlement Layer settlement model. Our primary authority on whether a place is an urban centre, town or rural grid cell.
Kontur Population An open dataset estimating residents inside neat hexagonal cells about 0.74 km² each. Our fallback classifier.
WorldPop A 100 m resolution population grid, UN-adjusted. Used to validate the other two.
Admin 1 / 2 / 3 Official geography: region, then district, then commune. The borders governments and programmes actually report against.
Displacement line A dashed line joining a stove's primary and secondary known locations: movement made visible.
Reasoning trail The recorded answer to "why is this record labelled Rural?" including source, threshold and date.

See the overlays working together

The interactive map and full dashboard are open now, no signup required.

Open the Interactive Map →