Climate Risk Indicators: What They Measure and Why They Matter

Three very different things get called “climate indicators,” and most confusion in this field starts with mixing them up. Global system indicators tell you the planet is warming. Country loss indices tell you which nations have been hit hardest in the past. Asset-level hazard indicators tell you whether a specific site faces rising flood, heat, or water risk over the next 30 years. Only the third kind can carry a screening decision, but you will meet all three in reports, frameworks, and vendor pitches, so it pays to know exactly what each one measures and where its authority ends.

This guide walks through all three from a practitioner’s seat: what the indicator is, who publishes it, what question it can answer, and the mistakes people make when they stretch it past that question.

The 7 global climate indicators (WMO), and what they cannot tell you

The canonical set of seven global indicators comes from the World Meteorological Organization (WMO), which selected them in 2018 to describe the state of the climate system in terms the public and decision-makers can track:

• Surface temperature
• Ocean heat content
• Atmospheric carbon dioxide concentration
• Ocean acidification
• Sea level
• Glacier mass balance
• Arctic and Antarctic sea ice extent

The WMO assesses each of these at least annually in its State of the Global Climate reports, and its recent reports show record or near-record values across the set. A related list you will also see cited is NOAA’s “Ten Signs of a Warming World,” a ten-indicator set that adds variables such as sea surface temperature, humidity, and snow cover.

Here is the practitioner’s point: these indicators diagnose the system, not your exposure. They confirm the direction and pace of change with high confidence, which makes them the right citation for the “why this matters” paragraph of a board deck. They cannot rank your sites, price your risk, or tell you whether a warehouse in Jakarta needs a flood study. Every one of them is a planetary or regional aggregate. The moment a decision attaches to a specific location, you need a different class of indicator.

The Climate Risk Index ranks past national losses, not your future risk

The Global Climate Risk Index (CRI), published by Germanwatch, ranks countries by the human and economic toll of extreme weather events that already happened. The current edition, CRI 2026 (published November 2025), draws on the EM-DAT international disaster database along with World Bank and IMF data, and ranks countries on fatalities and economic losses, in absolute terms and relative to population and GDP. Its long-term ranking for 1995 to 2024 names Dominica, Myanmar, and Honduras as the most affected countries. Publication paused for several years after the 2021 edition while the index moved off its previous insurance-industry data source, then resumed with the 2025 edition.

Read the CRI for what it is: a well-documented, backward-looking measure of realized national losses. It is useful context for country concentration discussions and for making the human cost of extreme weather concrete. It is the wrong tool for asset decisions, for three reasons a practitioner should be able to recite:

1. It measures the past. A country can rank low because a catastrophe has not happened yet, not because it will not.
2. It is national. A country-level rank says nothing about which of your sites inside that country carries the exposure.
3. It mixes hazard with vulnerability. Losses reflect building stock, preparedness, and wealth as much as weather, so a low rank can mean “resilient,” not “safe.”

For forward-looking, location-specific questions, practitioners use hazard indicators built from climate projection data.

The 12 asset-level indicators practitioners screen with

A location-specific physical climate risk assessment screens a defined set of hazards, each with an indicator you can measure, project, and defend. A common structure covers 12 hazards in four families. For each indicator below, note the unit: a defensible indicator always has one.

Temperature indicators

• Heat wave days per year. Days above a location-specific temperature threshold derived from the historical baseline. This is the workhorse for labor productivity, cooling load, and infrastructure stress questions. One caveat: threshold methods built on air temperature tend to understate heat risk in already-hot tropical climates, so treat tropical heat readings as a floor.
• Cold stress days per year. Days below a minimum-temperature threshold. Relevant for freeze-thaw damage and cold-chain operations in temperate and continental zones; it reads zero in the tropics, as it should.
• Mean temperature change (°C). The projected shift in average temperature against baseline. A context indicator: it tells you the trajectory, not an impact.

Precipitation indicators

• Drought months per year. Months of precipitation deficit against historical norms, in the spirit of a standardized precipitation index. Consecutive dry months are the signal that matters for water-dependent operations.
• Extreme rainfall days per year. Days exceeding rare-intensity rainfall thresholds. This is the indicator most tightly linked to flood risk, flash flooding, and drainage failure, and basic physics pushes it upward: warmer air holds roughly 7% more moisture per degree of warming.
• Annual precipitation change. The shift in total rainfall against baseline. Direction matters as much as size; gaining and losing regions face different risk profiles.

Compound indicators

• Fire weather days per year. An estimate of days with wildfire-prone conditions. Be careful with method claims here: full fire-weather indices need temperature, humidity, and wind together, and many screening-grade indicators are temperature-driven proxies adjusted for land cover (a city center or open water cannot carry wildfire regardless of temperature). Ask any provider which of the two they compute; both are legitimate as long as the method says which one it is.
• Landslide susceptibility. A rating built from slope, rainfall intensity, and terrain, not a daily count. Flat terrain should read low; steep wet terrain high. If it does not, question the terrain data.
• Severe storm exposure. Extreme wind frequency, ideally combined with the observed tropical cyclone track record near the site (NOAA’s IBTrACS archive is the reference source). Wind projections alone miss cyclone exposure; the historical track record alone misses change. You want both.

Hydrological indicators

• River flood exposure. Built from terrain, catchment position, and projected precipitation. In screening-grade methods, precipitation typically serves as the proxy for river discharge; full hydrological discharge modeling lives a level deeper, in dedicated flood models with mapped depths per return period (a “1-in-100-year” flood is one with a 1% annual chance).
• Sea level rise (metres). Projected local sea level change from the IPCC AR6 regional projections, read against site elevation. It should be marked not applicable for inland or high-elevation sites; a sea level number attached to a plateau city is a red flag about the method.
• Water stress ratio. Water demand divided by available supply for the basin, sourced from WRI Aqueduct. Ratios above 0.4 classify as high stress. This is a basin-scale indicator (basins average around 10,000 km²), so it describes the watershed you sit in, not your intake pipe.

Read the change from baseline, not the absolute number

A raw indicator value means little on its own. “23 heat wave days per year by 2050” is not a finding; the finding is the movement. A site going from 2 days to 15 is experiencing a fivefold regime change it has never adapted to. A site going from 40 to 55 is a hot place getting hotter, with (usually) infrastructure already built for heat. Same absolute increase, different risk.

So practitioners read indicators in three steps:

1. Baseline first. What does this location experience today? The baseline is the adaptation the site already has.
2. Change second. How far does the projection move from that baseline, per scenario and horizon?
3. Rating last. Convert the value to a tier using disclosed thresholds. A five-tier scale (Low, Moderate, High, Severe, Extreme) is the common convention, with each hazard rated on its own physical scale: days per year for heat, months for drought, metres of remaining elevation buffer for sea level rise. The thresholds differ because the physics differ, and any provider should show you the cutoffs.

Composite scores are for triage; read the top risk first

You will often see the 12 hazard ratings compressed into one composite score per location. Used correctly, a composite is a triage tool: it lets you sort a thousand sites and decide where to look first, and geographic logic (coastal sites weight sea level and storm; steep terrain weights landslide) makes the sort smarter than a flat average.

Two cautions before you rely on one:

• Compression hides extremes. A site with one Extreme hazard and eleven Low ones can score the same as a site that is Moderate across the board, and they need completely different responses. Always read the top hazard alongside the composite; we walk through the failure mode in how a composite score can hide your worst risk.
• No disclosure framework asks for a single score. TCFD-aligned and ISSB-aligned reporting asks for exposure by hazard, scenario, and time horizon, and for the amount and share of assets vulnerable to physical risk. The composite is your sorting device, not the disclosure.

Check the coverage behind a composite too: a score built on 10 or more hazards with data is more trustworthy than one built on 5 or 6, and a good provider reports that coverage with the score.

Check projections against the observed record

The step most screening programs skip: compare the forward-looking indicator with what has actually happened at the location. Public event archives now make this practical; global flood event datasets assembled from news and satellite records, along with landslide catalogs, let you ask “how often has this place actually flooded, and in which months?”

The cross-check runs both ways:

• Projection high, record empty. Elevated modeled flood risk but no recorded events nearby. Maybe the site drains well, maybe protection exists, maybe the model is reading the river valley next door. Investigate before you spend on adaptation.
• Projection modest, record loud. A site with a dozen recorded flood episodes and a moderate forward rating deserves escalation regardless of what the model says. The record is evidence; a model cannot argue with it.

Where observed rainfall data exists for past flood episodes, you can go one step further and estimate the rainfall level that historically floods the site, then ask whether projections push more days past that trigger. That converts “risk goes up” into a concrete, defensible sentence: the rainfall that flooded this site in the past becomes N% more frequent by 2050.

What makes an indicator decision-grade

When you evaluate a data provider, or build indicators yourself, hold each one to six properties:

• Measurable, with a unit. “Heat wave days per year” qualifies. “Heat risk: high” without a defined metric does not.
• Baseline-referenced. The indicator shows change against a stated historical period, not a bare absolute value.
• Multi-scenario. Reported under at least two emissions scenarios so the range is visible. In practice that means SSP2-4.5, the middle scenario (about 2.7°C of warming by 2100), and SSP5-8.5, the high scenario (about 4.4°C); a low, Paris-aligned scenario (SSP1-2.6, about 1.8°C) adds the third point where available.
• Documented and versioned. The data source, threshold definitions, and assumptions are written down, and the method carries a version so results can be reproduced and audits supported. If a provider cannot explain how an indicator is calculated, the indicator is unreliable, and if the method is not versioned, last year’s numbers cannot be defended this year.
• Resolution that matches the decision. For asset-level screening, projections downscaled to roughly 25 km, such as NASA’s NEX-GDDP-CMIP6, are the working standard; country averages are not screening data. Know the resolution of every layer you rely on, because a 25 km climate grid and a 30 m terrain model answer different questions.
• Confronted with the record. The provider can show you the observed event history next to the projection, per the section above.

One scope note: climate indicators are a subset of the broader environmental indicator families used in ESG reporting (air and water quality, biodiversity, land use). Those matter for other disclosures; physical climate risk screening runs on the climate subset.

How to choose indicators for an assessment

The selection method, in five steps:

1. Start from the asset and its geography. A coastal data center and an inland farm do not share a hazard list. Filter the 12 indicators to the ones your geography can physically deliver.
2. Fix scenarios and horizons before you look at results. The middle and high scenarios at 2030 and 2050 are the common frame for TCFD-aligned work. Deciding after you see the numbers invites cherry-picking.
3. Read change from baseline (the section above), per scenario, per horizon.
4. Use the composite to sort, the top hazard to decide, and check the data coverage behind every score.
5. Escalate the shortlist. Screening indicators identify which sites deserve deeper work: site flood studies, engineering surveys, financial modeling of the loss. Translating the shortlist into money terms (asset values, downtime, insurance structure) is where indicator work hands off to climate risk management as a process.

This is the same screen-then-escalate architecture the disclosure standards themselves expect: broad and consistent first, deep and site-specific where the screen says it matters.

Frequently Asked Questions

What is a climate indicator?

A measurable variable that tracks either the state of the climate system or climate-related risk at a location. Global indicators (surface temperature, sea level, ocean heat) monitor planetary trends. Location-specific indicators (heat wave days per year, drought months, flood exposure) quantify physical risk for individual assets and portfolios.

What are the 7 climate indicators?

The World Meteorological Organization’s seven global climate indicators: surface temperature, ocean heat content, atmospheric carbon dioxide, ocean acidification, sea level, glacier mass balance, and Arctic and Antarctic sea ice extent. NOAA separately maintains a ten-indicator set, the “Ten Signs of a Warming World,” which adds variables such as sea surface temperature, humidity, and snow cover.

What is the Climate Risk Index?

An index published by Germanwatch ranking countries by the human and economic losses from past extreme weather events, based on the EM-DAT disaster database with World Bank and IMF data. The 2026 edition’s long-term ranking (1995 to 2024) names Dominica, Myanmar, and Honduras as most affected. It measures realized national losses, not forward-looking or asset-level risk.

What are the three types of climate risks?

Physical risk (direct impacts from climate hazards), transition risk (financial effects of the shift to a low-carbon economy), and liability risk (legal claims arising from climate damages or disclosure failures). Most disclosure frameworks, including TCFD and IFRS S2, structure reporting around the first two, with liability risk often treated within transition risk.

What are climate metrics?

Quantitative measures used to assess and report climate performance or risk: emissions by scope, carbon intensity, physical hazard indicators, exposure amounts, and climate value-at-risk. “Climate metrics” tends to appear in financial reporting contexts; “climate risk indicators” emphasizes the underlying physical measurements. The two overlap heavily.

What is the biggest indicator of climate change?

Global surface temperature is the most cited. Ocean heat content is arguably the most telling: the ocean absorbs over 90% of the excess heat trapped by greenhouse gases, making it the least noisy measure of the planet’s energy imbalance.