What Geocoding Does
Geocoding is the process of converting a text address (such as “221B Baker Street, London”) into geographic coordinates (latitude and longitude). This is typically the first step in any Continuuiti analysis workflow — you geocode your addresses, then use the resulting coordinates to run LULC, Climate Risk, Flood Depth, or other analyses.
Continuuiti uses Nominatim, the geocoding engine built on OpenStreetMap (OSM) data. OpenStreetMap is a collaborative, community-maintained global map database — think of it as the Wikipedia of maps. Nominatim searches this database to find the best match for your address and returns the corresponding coordinates.
Nominatim was chosen because it provides global coverage at no per-query cost, uses openly licensed data, and avoids vendor lock-in. Coverage quality varies by region — it is excellent in Europe, North America, and urban areas worldwide, and less complete in rural areas of developing countries.
How Geocoding Quality is Measured
Each geocoded address receives three quality indicators: a quality score, a quality status, and a match type.
The quality score (0.0 to 1.0) is the Nominatim importance score, which reflects how well-known and precisely defined the matched location is. A score above 0.7 generally indicates a confident match to a specific address or building. Scores below 0.3 may indicate a vague match to a city or region.
The quality status tells you the outcome of the geocoding attempt:
- PASSED — The address was successfully geocoded. Coordinates are available.
- REJECTED — The address failed pre-validation quality checks and was not sent to Nominatim at all.
- FAILED — The address passed pre-validation but Nominatim could not find a match.
The match type tells you the precision level of the match:
| Match Type | Accuracy | Description |
|---|---|---|
| building | ~10 m | Matched to a specific building footprint |
| address | ~20 m | Matched to a specific house number on a street |
| street | ~100 m | Matched to a street, coordinates at street centroid |
| city | ~1 km | Matched to a city or suburb centroid |
| country | ~100 km | Matched to a country only |
Higher-precision match types (building, address) produce more reliable downstream analyses.
Pre-Validation Quality Checks
Before sending an address to Nominatim, Continuuiti runs five instant quality checks to detect obviously unusable input:
| Check | Triggers When | Example |
|---|---|---|
too_short |
Address is fewer than 5 characters | “NYC” |
no_alphanumeric |
Contains zero letters or numbers | “—///—“ |
suspicious_characters |
Contains emoji or characters like <>{} | “123 Main St” |
repeated_chars |
Same character appears 5+ times | “aaaaaaa” |
no_spaces |
No whitespace in addresses >10 chars | “123MainStreetNewYork” |
Addresses that fail any check are immediately rejected without consuming a Nominatim query or rate-limit slot.
Why Addresses Fail
Pre-validation rejection catches obviously malformed input — gibberish, extremely short strings, or text containing emoji or special characters. These are rejected instantly.
Nominatim lookup failure occurs when a well-formed address cannot be found. Common causes: the address does not exist in OpenStreetMap, uses local abbreviations Nominatim does not recognise, or is ambiguous between multiple locations.
Regional coverage gaps are a significant factor. OpenStreetMap coverage is built by volunteers, so urban areas in Europe and North America tend to have excellent address-level data, while rural areas in parts of Africa and South Asia may have limited coverage.
Fixing Rejected Addresses
When an address is rejected or fails, you can edit and resubmit at no additional credit cost. Retries are always free.
To improve success: provide complete addresses including street number, street name, city, postal code, and country. Avoid abbreviations. For locations where street addresses are unreliable, try using a well-known landmark name instead.
Bulk Geocoding
Bulk geocoding accepts up to 5,000 addresses in a single CSV upload. Addresses are processed sequentially because Nominatim enforces a fair-use rate limit of one request per second.
| Batch Size | Approximate Time |
|---|---|
| 100 addresses | ~2 minutes |
| 500 addresses | ~8 minutes |
| 1,000 addresses | ~17 minutes |
| 5,000 addresses | ~83 minutes |
Rejected addresses skip both the API call and the rate-limit delay, so batches with many invalid addresses complete faster. Results can be exported as CSV with status filtering.
Limitations
OpenStreetMap coverage varies significantly by region. Results are most reliable in Europe, North America, and major urban centres worldwide. Rural areas, informal settlements, and regions with limited OSM volunteer activity may have incomplete data.
Geocoded coordinates are approximate, not surveyed. Geocoding is not a substitute for parcel-level land surveying.
Continuuiti currently offers forward geocoding only (address to coordinates). Reverse geocoding (coordinates to address) is not available.
Address formats vary globally. Structured formats (number, street, city, postal code, country) geocode most reliably. Landmark-based or non-standard local conventions may produce lower-quality results.
Frequently Asked Questions
“Why is my batch taking so long?”
Nominatim enforces a rate limit of one request per second to ensure fair access. A 5,000-address batch requires approximately 83 minutes. This is a constraint of the Nominatim service, not a platform limitation.
“Why was my address rejected, not failed?”
Rejected means the address was caught by pre-validation checks before being sent to Nominatim — it was obviously malformed. Failed means the address passed validation but Nominatim could not find a match in its database.
“Can I geocode addresses in non-English languages?”
Yes. Nominatim supports multilingual OpenStreetMap data and can match addresses in local scripts (Arabic, Chinese, Devanagari, Cyrillic, etc.). Coverage quality varies by language and region.
Appropriate use: First-step address conversion for downstream analysis, batch portfolio geocoding, address quality assessment.
Not suitable for: Parcel-level legal boundary determination, official land surveying, or regulatory submissions requiring surveyed coordinates.