Banks screening loan portfolios. Insurers pricing policies for coastal properties. Supply chain managers evaluating whether a key supplier sits in a flood zone. All of them face the same question: how do you actually analyze physical climate risk for a specific location?
Most organizations either hire consultants who charge $50,000+ for a custom report or rely on generic portfolio-level ratings that hide site-specific variation. Neither approach scales to hundreds or thousands of locations.
This guide breaks down the methodology behind physical climate risk analysis, step by step. You will learn which data sources feed the models, how hazards are categorized and scored, how composite risk ratings are calculated, and what limitations to watch for when evaluating any provider’s methodology.
What Are the Types of Physical Climate Risk?
Physical climate risks fall into two categories based on how they manifest:
Acute risks are event-driven and sudden. Floods, wildfires, severe storms, and heat waves cause immediate damage to assets and operations. A single extreme rainfall event can shut down a manufacturing facility for weeks.
Chronic risks are gradual shifts that compound over decades. Rising average temperatures, changing precipitation patterns, sea level rise, and increasing water stress erode asset values and disrupt supply chains slowly but persistently.
A comprehensive physical climate risk analysis must account for both types. Focusing only on acute events misses the slow erosion of chronic changes. Focusing only on chronic trends misses the catastrophic tail risks. For a deeper breakdown, see our guide to physical climate risk assessment.
Step-by-Step: How to Analyze Physical Climate Risk
Whether you are building an internal capability or evaluating a third-party provider, every credible climate risk assessment methodology follows the same general framework. Here are the six core steps.
Step 1: Define Scope and Locations
Start by identifying the locations that matter. For a bank, that is every property in a mortgage portfolio. For a manufacturer, it is every facility, warehouse, and tier-1 supplier site. For a real estate fund, it is every asset under management.
Each location needs precise coordinates (latitude and longitude), not just city-level approximations. A property at sea level faces different risks than one 50 meters higher, even within the same city. Geocoding accuracy matters because the downstream hazard calculations depend on elevation, terrain, land cover, and proximity to coastlines.
Step 2: Select Data Sources
Physical climate risk analysis relies on climate projection datasets, not historical weather records. The standard approach uses global climate models (GCMs) from the Coupled Model Intercomparison Project, now in its sixth phase (CMIP6). These models simulate how the climate system responds to different levels of greenhouse gas emissions.
The most widely used dataset for location-level analysis is NASA NEX-GDDP-CMIP6, which provides bias-corrected, statistically downscaled projections at approximately 25 km resolution. It includes daily temperature (mean, max, min), precipitation, and surface wind speed variables from multiple climate models.
Supplementary data sources fill gaps that climate models cannot cover:
- WRI Aqueduct 4.0 provides basin-level water stress projections incorporating supply, demand, and groundwater trends
- IPCC AR6 supplies sea level rise projections with scenario-specific trajectories
- NASA SRTM (30-meter resolution) provides elevation and terrain slope for flood and landslide susceptibility
- ESA WorldCover (10-meter resolution) provides land cover classification for wildfire susceptibility
When evaluating a provider, ask which datasets they use. If they cannot name specific sources, that is a red flag.
Step 3: Choose Emission Scenarios
Climate projections are not predictions. They are “if-then” scenarios: if emissions follow this pathway, then temperatures, precipitation, and extreme events change by this much.
The current standard uses Shared Socioeconomic Pathways (SSPs) from the IPCC AR6 framework. Most analyses compare at least two scenarios:
- SSP2-4.5 (“Middle of the Road”): Moderate mitigation efforts, global warming of approximately 2.7°C by 2100. CO₂ peaks around 2050 then declines.
- SSP5-8.5 (“Fossil-Fueled Development”): High economic growth fueled by fossil energy, approximately 4.4°C warming by 2100. CO₂ increases continuously.
Running both scenarios shows the range of possible futures. For TCFD-aligned reporting, scenario comparison is a requirement, not an option. For more on how these pathways work, see our explainer on SSP scenarios.
Step 4: Assess Hazards Across Categories
A thorough physical climate risk analysis examines multiple hazard types, not just the obvious ones. Hazards are typically organized into four categories:
Temperature hazards include heat waves (days exceeding extreme thresholds), cold stress (days below safe operating temperatures), and overall temperature change from the historical baseline.
Precipitation hazards include drought (sustained precipitation deficits, often measured via the Standardized Precipitation Index), extreme rainfall events (P99 threshold exceedances), and long-term precipitation change.
Compound hazards depend on multiple variables interacting. Wildfire risk combines temperature, wind speed, and land cover type. Landslide risk combines extreme rainfall with terrain slope. Severe storms combine wind speed extremes with precipitation intensity.
Hydrological hazards include river flooding (driven by precipitation and catchment characteristics), sea level rise (driven by thermal expansion and ice melt, using IPCC projections), and water stress (driven by supply-demand imbalance at the basin level).
For each hazard, the analysis extracts climate variables for the selected scenarios and time horizons, then converts raw data into risk-relevant metrics. A location’s elevation, terrain type, land cover, and distance to coast all feed into the calculations as modifiers. See our full breakdown of climate hazard types and examples.
Step 5: Apply Risk Thresholds and Rating Scales
Raw metrics (heat wave days per year, drought months, flood discharge days) need to be translated into actionable risk ratings. The standard approach uses a tiered rating scale, typically five levels: Low, Moderate, High, Severe, and Extreme.
Each hazard has its own threshold breakpoints calibrated to the metric it measures. For example, a location experiencing fewer than a handful of heat wave days per year might rate Low, while one with dozens of extreme heat days would rate Severe or Extreme.
Good methodology is transparent about where thresholds come from. Some are grounded in published climate science (the Clausius-Clapeyron relationship for extreme rainfall scaling, IPCC AR6 for sea level rise). Others involve engineering judgment (terrain slope factors for landslide susceptibility). Any methodology that presents thresholds as black-box outputs without rationale deserves skepticism.
Step 6: Calculate Composite Risk Scores
Individual hazard ratings are useful but incomplete. A location might face Low flood risk yet Extreme heat wave risk. Decision-makers need a single composite score that weights and aggregates all hazards into an overall risk profile.
The standard approach uses a weighted average formula:
Composite Score = Sum(Score × Weight × Modifier) / Sum(Weight × Modifier)
Not all hazards deserve equal weight. Primary climate signals (heat waves, drought, flooding) typically carry higher weight than secondary or compound hazards (landslide, precipitation change). Context indicators that measure general trends rather than specific risks often receive reduced weight.
Geography matters too. Coastal locations amplify the weight of sea level rise and storm surge while reducing wildfire risk (maritime humidity). Inland locations exclude sea level rise entirely from the composite calculation.
The result is a single score per scenario and time horizon, mapped to the same five-tier scale. A confidence indicator based on data availability (what percentage of hazards returned valid results) accompanies every composite score.
Physical Climate Risk Scoring: What to Evaluate in a Provider
When comparing climate risk providers, the scoring methodology reveals more about quality than marketing materials. Ask these questions:
How many hazards are assessed? Comprehensive analyses cover 10-12 distinct hazards. If a provider only covers flood and heat, you are missing wildfire, landslide, water stress, and other significant exposures.
Are thresholds disclosed? Providers who publish their rating scales and threshold logic demonstrate confidence in their methodology. Black-box scores cannot be validated or challenged.
Does the composite score account for geography? A flat average across all hazards ignores the reality that coastal and inland locations face fundamentally different risk profiles. Geographic weighting adjustments separate rigorous methodologies from simplistic ones.
What confidence level is reported? Not every location has complete data for every hazard. Honest providers flag when data gaps reduce the reliability of the assessment.
Are limitations documented? Every climate projection has limitations. If a provider claims their analysis has none, they either do not understand their own methodology or are not being transparent.
Request a Sample Climate Risk Report
See how 12 physical hazards are scored for a real location. Free, no commitment.
Data Sources for Physical Climate Risk Analysis
The credibility of any analysis depends on its data inputs. Here are the primary data sources used across the industry:
| Data Source | Provider | Resolution | Used For |
|---|---|---|---|
| NEX-GDDP-CMIP6 | NASA | ~25 km | Temperature, precipitation, wind projections |
| WRI Aqueduct 4.0 | World Resources Institute | Basin-level | Water stress projections |
| IPCC AR6 WG1 | IPCC | Global | Sea level rise trajectories |
| NASA SRTM | NASA/USGS | 30 m | Elevation, terrain slope |
| ESA WorldCover | European Space Agency | 10 m | Land cover classification |
The key variable from NEX-GDDP-CMIP6 is that it provides bias-corrected projections, not raw model output. Bias correction adjusts climate model outputs to better match observed historical patterns, reducing systematic errors that would otherwise distort location-level risk assessments.
For a detailed walkthrough of climate projection datasets and how to access them, see our guide to climate data APIs.
Time Horizons: Baseline Through 2050
Physical climate risk analysis typically covers four time horizons:
- Baseline (historical): Usually 1980-2014, representing observed climate conditions against which future changes are measured
- 2030 (near-term): A 20-year window centered on 2030, showing risks already locked in by current emissions
- 2040 (medium-term): The horizon where SSP2-4.5 and SSP5-8.5 begin to diverge noticeably
- 2050 (long-term): Where scenario differences become pronounced, revealing the consequences of policy choices made today
The 20-year averaging windows smooth out year-to-year natural variability, producing stable signals that reflect the underlying climate trend rather than random fluctuations. For TCFD reporting, showing risk evolution across at least two time horizons under different scenarios is expected.
Limitations Every Analysis Should Disclose
No physical climate risk analysis is perfect. Transparency about limitations separates trustworthy providers from those selling certainty they cannot deliver.
Resolution constraints. Climate projections at 25 km resolution represent a regional average, not site-specific microclimate. A factory on a hilltop and one in a valley floor within the same 25 km grid cell will experience different flooding risks that the model cannot distinguish.
Single-model vs. ensemble. Running projections from one climate model is computationally efficient but captures only one trajectory. Multi-model ensembles provide uncertainty ranges but require significantly more processing power. Know which approach your provider uses.
Missing variables. Humidity data is unreliable across CMIP6 models, which means wildfire assessments often rely on temperature-only proxies rather than the full Fire Weather Index. Runoff data is similarly unavailable in many datasets, forcing river flood estimates to use precipitation as a proxy.
No adaptation assumptions. Most analyses project future risk assuming no additional adaptation measures are implemented. In reality, cities build flood defenses, farmers shift crop varieties, and buildings get retrofitted. These projections show exposure, not residual risk after mitigation.
Not insurance-grade. Screening-level risk tiers (Low through Extreme) are designed for portfolio screening, due diligence, and regulatory disclosure. They are not calibrated for actuarial loss estimation or engineering design. High-risk ratings warrant further investigation with site-specific studies.
Continuuiti’s climate risk assessment covers 12 physical hazards across multiple scenarios and time horizons, with each report documenting the data sources, confidence levels, and methodological limitations specific to that location.
When to Use Physical Climate Risk Analysis
Physical climate risk analysis is appropriate for screening and strategic decisions:
| Suitable For | Not Suitable For |
|---|---|
| Portfolio screening and prioritization | Engineering design specifications |
| Acquisition due diligence | Insurance underwriting (actuarial pricing) |
| TCFD scenario-based disclosure | Legal or regulatory compliance certification |
| Strategic planning and resource allocation | Real-time operational decisions |
| Stakeholder education and awareness | Specific event prediction or loss quantification |
Understanding the appropriate scope prevents both underuse (ignoring climate risk in major decisions) and overuse (treating screening results as engineering-grade precision).
Frequently Asked Questions
What does physical climate mean?
Physical climate refers to the measurable aspects of climate that directly affect the physical environment, including temperature, precipitation, wind patterns, and sea level. In risk analysis, physical climate risk specifically covers threats from climate-driven events (acute hazards like floods and heat waves) and long-term shifts (chronic hazards like rising temperatures and water stress) that can damage assets and disrupt operations.
What is a physical climate risk assessment?
A physical climate risk assessment evaluates how climate-driven hazards affect a specific location or portfolio of locations. It uses climate projection data, emission scenarios, and risk thresholds to rate exposure from Low to Extreme across multiple hazards and time horizons. The output supports TCFD disclosure, due diligence, and strategic planning.
What are the 4 types of climate risk?
Climate risk divides into physical risk (damage from climate events and shifts) and transition risk (financial impacts from policy changes and market reactions). Physical risk further splits into acute (sudden events like storms and wildfires) and chronic (gradual changes like temperature rise and sea level rise). Some frameworks add liability risk and reputational risk as additional categories.
How many hazards should a climate risk analysis cover?
A comprehensive analysis covers 10 to 12 distinct hazards across four categories: temperature, precipitation, compound, and hydrological. Analyses covering fewer than 8 hazards risk missing significant exposures that could affect financial decisions.
What is the difference between physical and transition climate risk?
Physical risk comes from climate-driven events and changes that damage assets, such as flooding or heat stress. Transition risk comes from economic shifts associated with decarbonization, including carbon pricing and stranded assets. Both affect financial performance through different mechanisms and timelines. For a deeper comparison, see our guide to climate risk modeling.
What data sources are used for physical climate risk analysis?
The primary source is NASA NEX-GDDP-CMIP6 for bias-corrected climate projections. Supplementary sources include WRI Aqueduct 4.0 (water stress), IPCC AR6 (sea level rise), NASA SRTM (elevation), and ESA WorldCover (land cover). Together, these enable multi-hazard analysis across emission scenarios and time horizons.
