# Location Classification Methodology

**Version:** 1.0  
**Standard:** DEGURBA (UN Statistical Commission, March 2020)  
**Last Updated:** February 2026

---

## 1. Overview

This document describes the methodology used to classify customer locations as **Urban**, **Peri-Urban**, or **Rural**. The approach leverages pre-computed DEGURBA-compliant data from the Global Human Settlement Layer (GHSL).

---

## 2. Terminology Glossary

| Our Term | DEGURBA Official Term | Density Threshold | Cluster Population |
|----------|----------------------|-------------------|-------------------|
| **Urban** | Urban Centre | ≥1,500 people/km² | ≥50,000 |
| **Peri-Urban** | Towns & Semi-dense Areas (Urban Cluster) | ≥300 people/km² | ≥5,000 |
| **Rural** | Rural Grid Cells | <300 people/km² | N/A |

> **Note**: We use "Peri-Urban" for user-friendliness. This maps to DEGURBA's "Towns and Semi-dense Areas" category.

---

## 3. Data Source Priority

Classification uses a **priority cascade** (not triangulation):

| Priority | Source | Resolution | What It Provides |
|----------|--------|------------|------------------|
| 1️⃣ Primary | **GHSL SMOD** | 1 km grid | Pre-computed DEGURBA classification with cluster analysis |
| 2️⃣ Fallback | **Kontur Population** | H3 Res 8 (~0.737 km²) | Population density per hexagon |
| 3️⃣ Last Resort | **WorldPop** | 100m grid | Population count per pixel |

### Why GHSL SMOD is Primary

GHSL SMOD codes are the **output of full DEGURBA implementation** by the EU Joint Research Centre. The codes already incorporate:

- ✅ Density threshold filtering (300/km² and 1,500/km²)
- ✅ Contiguity testing (identifying connected grid cells)
- ✅ Cluster population validation (5,000 and 50,000 thresholds)
- ✅ Gap-filling and edge smoothing

**We do not need to implement cluster analysis** because GHSL SMOD has already done it correctly.

---

## 4. GHSL SMOD Classification Codes

| Code | GHSL Name | Our Category |
|------|-----------|--------------|
| 30 | Urban Centre | Urban |
| 23 | Dense Urban Cluster | Urban |
| 22 | Semi-Dense Urban Cluster | Peri-Urban |
| 21 | Suburban | Peri-Urban |
| 13 | Rural Cluster | Rural |
| 12 | Low Density Rural | Rural |
| 11 | Very Low Density Rural | Rural |
| 10 | Water | N/A |

---

## 5. Kontur Density Thresholds (Fallback)

When GHSL SMOD is unavailable, we use Kontur population density:

| Classification | People per H3 Res 8 Hexagon | Equivalent Density |
|---------------|-----------------------------|--------------------|
| Urban | ≥1,000 | ~1,356 people/km² |
| Peri-Urban | 300–999 | ~407–1,355 people/km² |
| Rural | <300 | <407 people/km² |

### H3 Resolution 8 Specifications
- **Average area**: 0.737327 km² (737,327 m²)
- **Average edge length**: 461 m
- **Note**: "400m hexagon" is Kontur's shorthand, not actual dimensions

---

## 6. Compliance Statement

This methodology is **DEGURBA-compliant** because:

1. Primary data source (GHSL SMOD) is produced by EU JRC using official DEGURBA methodology
2. SMOD codes incorporate cluster identification and population thresholds
3. Fallback sources use density thresholds derived from DEGURBA standards

---

## 7. References

- [DEGURBA – A New Degree of Urbanisation](https://ec.europa.eu/eurostat/web/degree-of-urbanisation) (Eurostat)
- [GHSL Settlement Model](https://ghsl.jrc.ec.europa.eu/ghs_smod2019.php) (EU JRC)
- [H3 Resolution Table](https://h3geo.org/docs/core-library/restable/) (Uber H3)
- [UN Statistical Commission Report](https://unstats.un.org/unsd/statcom/51st-session/) (March 2020)