How insurers use climate risk maps: from quote to accumulation
Insurers use climate risk maps at four points: at quote time, to see which hazards affect a single location and how strongly compared with the rest of the country; in accumulation management, to see where policies pile up in the same hazard zone; at renewal, to decide which part of the book needs new terms or limits; and in risk engineering, to decide which sites deserve a survey first. A map layer answers “where and how much relative to elsewhere”; it does not price the risk. Pricing and capital need two further numbers, average annual loss and probable maximum loss, which are built on top of the hazard layer once exposure and vulnerability are added.
What a climate risk map is, and what it is not
A climate risk map is a spatial layer that gives every location a value for one hazard: flood, hail, lightning, storm wind, heatwave and so on. Three things need to be kept apart.
| Term | What it tells you | What it does not tell you |
|---|---|---|
| Hazard map | How strongly a peril affects each location, usually as a 1-100 score relative to the modelled country | Whether a building will be damaged, or how much the damage costs |
| Risk map | The hazard combined with what is exposed and how vulnerable it is; often shown as expected loss or damage ratio per location | A single event outcome; it is an average over many possible years |
| Loss metric (AAL, PML) | A monetary figure for pricing (AAL) or for limits, reinsurance and capital (PML at a return period) | Where the risk is; the metric is a number, the map is the picture behind it |
Definitions used on this page:
- Hazard score: a 1-100, country-relative index for one peril within the modelled geography. It compares and ranks locations. It is not a loss ratio and not a probability of loss.
- Average annual loss (AAL): the expected loss per year averaged over the long run; the climate component of the pure premium.
- Probable maximum loss (PML): the severe-but-plausible loss at a chosen return period; used for limits, reinsurance structuring and capital.
- Return period: a probability statement. A 100-year event has roughly a 1 percent chance of being exceeded in any given year.
- Accumulation: the part of a portfolio that would be hit by the same event.
Four decisions, one map layer
A map earns its place when it changes a decision. In insurance there are four.
| Decision | Question the map answers | Who uses it | Typical output |
|---|---|---|---|
| Quote time | Which perils affect this address, and how strongly compared with the rest of the country? | Underwriter, broker | Score per peril for one coordinate; accept, decline or add terms |
| Accumulation | Where do policies cluster in the same hazard zone? | Portfolio and reinsurance team | Hotspot map, concentration by zone, input to PML |
| Renewal | Which part of the book needs new terms, limits or a different appetite? | Underwriting management | Segmentation of the book by hazard class |
| Survey priority | Which sites deserve a risk-engineering visit first? | Risk engineering | Ranked site list from the map |
Quote time
At quote time the underwriter has an address or a coordinate and a few minutes. The useful answer is a per-peril score for that exact point, not a postcode average, delivered either on screen or by an API call that returns in real time. Two buildings on the same street can carry different flood scores; the map resolves that difference only if its resolution is finer than the distance between them.
Accumulation
Accumulation is where a map outperforms a table. Plotting the whole book on the hazard layer shows where policies stack up in the same flood plain, hail corridor or wildfire interface. The same picture feeds probable maximum loss at the chosen return period and, from there, reinsurance structure and capital.
Renewal
At renewal the question moves from one location to the segment: which hazard classes have been growing in the book, where the appetite thresholds should move, and which contracts need new terms or limits. A hazard layer with consistent national scaling lets a renewal team segment the book by class rather than by anecdote.
Survey priority
Risk engineering budgets are finite. A ranked list from the hazard map (highest score, highest exposure first) decides which sites get a visit this year.
Hazard score versus AAL and PML
A hazard score and a loss metric answer different questions, and mixing them is the most common mistake in climate risk reporting.
| Hazard score | AAL | PML | |
|---|---|---|---|
| Unit | 1-100, country-relative | Currency per year | Currency at a return period |
| Needs exposure and vulnerability? | No | Yes | Yes |
| Use | Compare and rank locations; quote-time terms; survey priority | Pricing, provisioning | Limits, reinsurance, capital |
| Can two locations with the same score have different values? | Not applicable | Yes, if the buildings differ in value or vulnerability | Yes |
The chain from hazard to financial impact, including PD, LGD and DSCR for lenders, is explained on physical climate risk: from hazard to financial impact.
Data quality questions to ask any provider
Before a map goes into an underwriting system, these questions decide whether it is usable. None has a universal right answer; the point is to have an explicit one.
- Resolution per peril. What is the cell size, and is it finer than the distance between neighbouring risks? Meteorological perils and flood usually have different resolutions; ask for each.
- Scale. Is the score country-relative or absolute? A country-relative 80 in two different countries is not the same hazard.
- Reference period. Which years does the hazard climatology cover, and when was it last recomputed?
- Validation. Has the layer been checked against recorded events, and are the results published?
- Forward-looking layers. Which emission scenarios and horizons are available, and at what resolution?
- Formats. Can the layer be delivered as a raster for your own GIS, as an API, as a batch file and as a report, so that quote time and portfolio work use the same data?
- Integration. How many locations can one batch handle, and what is the response time at quote time?
- Method transparency. Is the modelling documented, and can the provider explain why a given location scores as it does?
- New geographies. How long does a new country model take?
- Versioning. Are layer versions dated, so that a renewal can be compared with last year’s assessment?
What UrClimate Score delivers
UrClimate Score delivers each hazard as a map layer at location level in four coverage tiers: Türkiye with the full layer stack; the UK, Germany and Northern Cyprus at high resolution; 36 European markets through a regional package; and 182 countries by global screening. For Türkiye the layers are the ten meteorological perils (hail, lightning, storm wind, severe precipitation and flash flood, snow load, frost, cold air, fire weather, heatwave, drought), each scored 1-100, plus topographic flood at 25 m, landslide at 100 m, earthquake at about 500 m, wildfire at building level, and flood depth with average annual loss at 30 m for six return periods from 5 to 250 years. Earthquake AAL, PML and net loss form a separately licensed premium layer. Future climate metrics cover SSP2-4.5 and SSP5-8.5 for 2030, 2050 and 2100.
Layers are delivered as interactive maps in the dashboard, as PDF reports, as GeoTIFF rasters for your own GIS and through the API, with up to 100,000 locations per batch. The flood score was checked against 45 real flood events reported in Türkiye between 2014 and 2024; every event location scored between 62 and 94 out of 100, none below 60. The lightning method was published at EGU 2020. The API has been live since 2021. A new country model takes 1 to 4 months. Coverage tiers and resolutions are listed on the UrClimate Score page; the insurance workflow from quote to portfolio report is on the insurance and reinsurance page.
Frequently asked questions
Is a hazard map the same as a risk map?
No. A hazard map shows how strongly a peril affects each location, usually as a country-relative score. A risk map adds what is exposed and how vulnerable it is, and is usually expressed as expected loss or a damage ratio.
Can an underwriter price a policy from a hazard score alone?
No. A score compares and ranks locations and supports accept-and-terms decisions. Pricing needs average annual loss, which requires exposure and vulnerability on top of the hazard.
At which points in the insurance workflow does a climate risk map enter?
At quote time (per-location check), in accumulation management (where policies cluster), at renewal (segmentation and appetite) and in risk engineering (survey priority).
What resolution does an underwriting map need?
Finer than the distance between neighbouring risks. For Türkiye, UrClimate Score provides topographic flood at 25 m, landslide at 100 m, earthquake at about 500 m and meteorological perils from 3 km to 1 km.
How is a map delivered into an underwriting system?
As an API query with latitude and longitude at quote time, as a periodic batch file, as GeoTIFF rasters for the insurer’s own GIS, or as map layers and PDF reports in the dashboard.
Are portfolios outside Türkiye covered?
Yes. Global screening covers 182 countries, with the deepest dataset across Türkiye and the surrounding region; the UK, Germany and Northern Cyprus are covered at high resolution and 36 European markets through a regional package.
See your book on the map
A short first call to agree perils, geography and delivery format.
Write to us by email Request a demo
