---
title: "the watershed isn't in the asset register"
canonical_url: https://ensurance.app/guide/the-watershed-isnt-in-the-asset-register
markdown_url: https://ensurance.app/guide/the-watershed-isnt-in-the-asset-register.md
subtitle: "a climate risk model needs hazards, assets and damage curves. the ground that sets the hazard has no row"
category: natural-capital
---

# the watershed isn't in the asset register

*a climate risk model needs hazards, assets and damage curves. the ground that sets the hazard has no row*

Picture a water utility's risk register. The intake, the treatment plant, the pumps and the substation each have a location, an owner and a replacement value, and a score for flood, fire and heat. The burned forest above the intake has none of those. It is not on the list, so no model run against the list can lose it or credit it.

**What is a climate risk model?** A climate risk model estimates what physical hazards could cost by combining three inputs: the hazard (how often and how hard floods, fires or storms strike), the exposed assets and their value, and their vulnerability (how much damage each level of hazard does). It returns a loss for an average year and a bad one.

That is the model's whole world. Above it, the watershed holds the snow and the slope, the forest holds the soil after rain, and the floodplain spreads the flood before it reaches the plant. They were setting the hazard long before anyone scored the plant, and they do it whether or not anyone pays for them. [**Ensurance**](/manual/ensurance?from=guide) is how the people who depend on that ground can pay for its keeping. It is not the watershed, and it does not stop a storm.

## what a climate risk model is made of

Every loss estimate, from a free public index to a reinsurer's catastrophe model, runs on three inputs. A hazard score is the first of them on its own. Catastrophe-model mechanics are in [the model learns from a past that is leaving](/guide/the-model-learns-from-a-past-that-is-leaving?from=guide).

| input | what it holds | where it usually comes from |
|---|---|---|
| Hazard | Flood depth, wind speed, flame length, heat, each with a likelihood, today and under future climate paths | Hazard maps, simulated events, climate projections |
| Exposure | An asset list: buildings, plants, roads, lines and pipes, with location, construction, value and owner | The client's own register, property records, building footprints |
| Vulnerability | Damage curves: the share of an asset's value lost at each level of hazard, by type of structure | Engineering tests, past claims, published curves |

Run the events against the assets and the model returns **expected annual loss** (EAL), the long-run average cost per year, and a bad-year figure such as **probable maximum loss** (PML), the loss exceeded about once in 100 or 250 years. Which of the two a decision should lean on is [the second post in this series](/guide/the-average-year-is-not-the-year-that-breaks-you?from=guide).

The exposure input is an asset register, built from what the client owns, insures or lends against. That is reasonable: the client's loss lands on the client's assets.

## the row that isn't there

Now walk uphill from the plant. The watershed above the intake, the floodplain upstream of the town and the marsh on the coast each change the hazard, or the vulnerability, before it reaches anything on the register. A burned slope sends sediment and debris into the intake. A filled floodplain sends deeper water downstream. A narrower marsh lets more wave energy through.

Few of them fit an input cleanly. The watershed is not an exposed asset: the utility does not own it, cannot insure it and has no replacement value to enter. It is not a damage curve; it changes what reaches the things that are damaged. Where the ground shows up at all, it is usually folded into the hazard layer as terrain or a land-cover class: scenery, not something with an owner that someone could pay to keep.

The model can give the plant an expected annual loss from flood or sediment. It cannot say whose land sets that number, what condition that land is in, or what happens to the loss if someone pays to keep it.

:::johnson
**a risk model can only lose what it has a row for.**

The hazard, the asset list and the damage curves describe the plant. The ground that sets much of its hazard sits between the rows, owned by someone else and, as a thing someone could pay to keep, counted by almost no one.

[see the Risk Resilience flow →](/risk-resilience.ensurance?from=guide)
:::

## where nature already shows up

Existing natural defenses can sit inside the hazard. At least one commercial flood model's public documentation lists existing wetlands, mangroves and floodplain restoration among the adaptations it accounts for: the marsh that is there today, as part of the ground the water crosses.

Researchers have also run living systems through industry-grade models: the same events, with the ground and without it.

| study | what was modeled | result | kind of figure |
|---|---|---|---|
| Narayan et al., 2017 | Coastal wetlands in Hurricane Sandy, on industry catastrophe models | About $625M of direct damage avoided across the northeastern US; in Ocean County, New Jersey, annual flood losses 16% lower on average, up to 70% lower at low elevation | Modeled |
| The Nature Conservancy and WTW, 2021 | Ecological forestry in the Northern Sierra, in catastrophe models | Average annual loss 40–60% lower | Modeled |
| Swiss Re Institute, 2025 | National Flood Insurance Program (NFIP) claims in Florida, 2009–2022, against coastal habitat | Best-protected areas filed about half as many claims in frequent, less severe storms | Observed, claim frequency only |

The ground can be modeled. Each study was built for one mechanism in one kind of place.

## where it does not

What is still missing is the part a decision needs.

**A proposed project upstream.** Counting the marsh that exists is one job. Pricing a thinning above a reservoir or a reconnected floodplain on a ranch, on land the client does not own, is a different one, and it needs its own evidence case by case.

**The owner.** The ground above the intake usually belongs to a forest agency, a ranch, a Tribal nation or a land trust, not to the party carrying the loss.

**Condition over time.** A land-cover class set years ago cannot tell a thinned forest from an overgrown one.

**The dependents.** One watershed can feed an intake, a highway and a town. A model run for one client sees one of them.

Pricing has the same gap. The Environmental Defense Fund wrote in July 2026 that NFIP pricing does not "clearly or consistently" credit nature and that catastrophe models are "not regularly updated", and proposed a post-model adjustment for nature-based solutions. A fix applied after the model is a sign the row was never in it. EDF adds that lower premiums alone will not drive investment in nature, and Resources for the Future found in 2025 that wildfire retrofit costs are "orders of magnitude greater than the insurance savings." A premium credit rarely pays for the work.

## what a row for the ground would need

To sit beside the plant in a decision, the ground needs what any row has.

| field | what it holds | why the loss model needs it |
|---|---|---|
| Location | The named parcels or reach of the watershed, mapped | So the hazard specialist knows which slope, channel or marsh edge to model |
| Owner | Who holds the land, and who looks after it | So there is someone to pay and to agree the work with |
| Condition | Fuel, cover, channel and marsh edge, read the same way each year | So each run has a current input, not an old class |
| Mechanism | What it changes (runoff, sediment, debris, fuel, surge) and whether that is a hazard or a vulnerability term | So the analyst changes the right term, not an assumed percentage |
| Dependents | The intakes, plants, roads and towns below it | So the avoided loss lands on named budgets, and the people who carry it can share the cost |

That is a record, not a model, and it is the record **ensurance** is built to keep. A certificate can name the parcels. An account in the place's name can hold what the dependents pay, and proceeds go to the people doing the keeping. Permanence is a path: a certificate moves from ensured toward entrust, permanent protection. The flows the protocol tracks for this work sit under [Risk Resilience](/risk-resilience.ensurance?from=guide).

Finding the ground is a different job from holding it: [who sells nature and climate location data](/guide/who-sells-nature-location-data?from=guide). One percentage for one service in a hard year is another tool again ([what nature value at risk actually measures](/guide/what-nature-value-at-risk-actually-measures?from=guide)). The portfolio version of the question is [long or short the source?](/guide/long-the-source?from=guide)

## the honest limit

A row is not a result. Turning the ground into a loss figure still takes a hazard specialist and a loss model, case by case, running the same events with and without the work. The method is in [how an intervention lowers the loss](/guide/how-an-intervention-lowers-the-loss?from=guide).

Public tools cover parts of that chain, and only parts. United States Geological Survey (USGS) post-fire debris-flow assessments estimate likelihood and volume under rainfall scenarios, not how well a treatment works or what a buyer saves. The Natural Capital Project's InVEST sediment model leaves out landslides, so it is no substitute for a debris-flow model.

The ground does not buy back the extreme tail. It mostly moves frequent and mid-size losses, often secondary perils such as wildfire and flood at a town's edge. A record event can exceed any buffer, and peak-peril capital stays put ([what reinsurance cannot cede](/guide/what-reinsurance-cannot-cede?from=guide)).

And **ensurance** does not sell a risk model or run catastrophe models. Supplier and public hazard data are inputs we read beside the ground.

## what to do with the register you have

Take your asset list and walk uphill. For every intake, plant, road or line, write down the ground that changes its hazard, and who owns it. That is not a loss figure. It is the rows your model has been missing.

For a US place's expected annual loss by peril from the Federal Emergency Management Agency (FEMA), start at [place](/place?from=guide); to see the living systems under it, [explore](/explore?from=guide). Once the ground has a row, the next question is when to pay for it: [waiting is a position](/guide/waiting-is-a-position?from=guide).

## frequently asked questions

### what is a climate risk model?

A model that estimates the cost of physical hazards from three inputs: the hazard, the exposed assets and their vulnerability to damage. It returns an average-year and a bad-year loss for the assets on its list.

### do climate risk models include nature?

Sometimes, as part of the hazard. At least one commercial flood model counts existing wetlands and mangroves, and studies have run wetlands and forest treatment through catastrophe models. What they rarely hold is a proposed project on someone else's land, its owner, its condition over time and the dependents below it.

## sources

[Narayan, S. et al. (2017), "The value of coastal wetlands for flood damage reduction in the northeastern USA," *Scientific Reports* 7: 9463](https://doi.org/10.1038/s41598-017-09269-z) — About $625M of direct damage avoided in Hurricane Sandy; Ocean County, New Jersey, annual flood losses 16% lower on average and up to 70% lower at low elevation; modeled on industry catastrophe models

[The Nature Conservancy and WTW (2021), *Wildfire Resilience Insurance*](https://wtwco.com/en-US/Insights/2021/07/wildfire-resilience-insurance-quantifying-the-risk-reduction-of-ecological-forestry-with-insurance) — Ecological forestry in the Northern Sierra lowered modeled average annual loss 40–60%

[Swiss Re Institute (2025), "Natural habitats can reduce flood losses"](https://www.swissre.com/institute/research/topics-and-risk-dialogues/climate-and-natural-catastrophe-risk/coastal-and-flood-protection-natural-habitats.html) — Florida NFIP claims, 2009–2022: about half the claim frequency in the best-protected areas in frequent, less severe storms

[A commercial flood model's public adaptation documentation](https://help.firststreet.org/hc/en-us/articles/360050275794-The-Nation-s-first-flood-adaptation-database) — Existing natural protections, including wetlands, mangroves and floodplain restoration, among the adaptations counted

[Environmental Defense Fund (July 2026), "Bringing nature into the NFIP"](https://blogs.edf.org/markets/2026/07/28/bringing-nature-into-the-nfip/) — NFIP pricing does not "clearly or consistently" credit nature; catastrophe models "not regularly updated"; post-model adjustment for nature-based solutions among four recommendations; lower premiums alone will not drive investment in nature

[Resources for the Future, Working Paper 25-30, *From Risk to Reward* (2025)](https://www.rff.org/publications/working-papers/from-risk-to-reward-insurance-discounts-for-wildfire-mitigation/) — Wildfire property retrofit costs are "orders of magnitude greater than the insurance savings"

[USGS, post-fire debris-flow hazard assessment](https://www.usgs.gov/programs/landslide-hazards/science/postfire-debris-flow-hazard-assessment-data) — Likelihood and volume of debris flows under rainfall scenarios

[Natural Capital Project, InVEST sediment delivery ratio model, user guide](https://storage.googleapis.com/releases.naturalcapitalproject.org/invest/3.20.0/userguide/en/sdr.html) — Scenario-based sediment modeling; landslides and other sediment sources are outside its scope

## the series

1. [the score is the easy half](/guide/the-score-is-the-easy-half?from=guide) — **Read first:** What a climate risk score leaves undecided, and the five questions after it
2. [the average year is not the year that breaks you](/guide/the-average-year-is-not-the-year-that-breaks-you?from=guide) — Expected annual loss is an average; the bad year sets the capital, and living systems move some of it, not all
3. [the bet that isn't in your scenarios](/guide/the-bet-that-isnt-in-your-scenarios?from=guide) — A stress test reports the risk you modeled; the source your site depends on is usually the scenario nobody ran
4. [the watershed isn't in the asset register](/guide/the-watershed-isnt-in-the-asset-register?from=guide) — A climate risk model needs hazards, assets and damage curves; the ground that sets the hazard has no row
5. [waiting is a position](/guide/waiting-is-a-position?from=guide) — When the bad years get worse, the protection is worth more and each year of waiting costs more
6. [the cost-benefit case for the ground upstream](/guide/the-cost-benefit-case-for-the-ground-upstream?from=guide) — A worked example on a hypothetical water utility, from the loss to a payment a budget can book