Model Catalog / climate

NOAA GFDL Physical Risk Lens

Climate Current active

NOAA's process-based earth system models (CM4 and ESM4) — the physical process credibility anchor in CE's climate model library. Uniquely strong in coupled ocean-atmosphere dynamics (tropical cyclone intensification, storm surge), high-resolution hydrology (river discharge, groundwater recharge, drought intensity), and ice-sheet-informed sea level rise projections beyond 2050. CE uses GFDL to constrain CMIP6 ensemble tails and validate compound event probabilities — the reference model for insurance catastrophe tail calibration and long-duration infrastructure asset risk in hydrology-sensitive sectors and coastal geographies.

Horizon 2025–2100
Geography Global (high-resolution ocean-atmosphere coupling)
Resolution Process-level earth system; sector via hazard overlay
Projection years 2030, 2040, 2050, 2060, 2080, 2100
0.74
hazard
0.52
transition
0.44
resilience
0.75
confidence
Tropical cyclone intensity Compound multi-hazard events Hydrological cycle Sea-level rise (ice dynamics) Ocean heat content Coastal flooding Drought intensity
Ensemble breadth vs. process depth — why GFDL is the tail-risk anchor

CMIP6 achieves its breadth through diversity: 40+ models with different structural assumptions produce a probability distribution that captures the full range of possible outcomes. GFDL achieves its value through depth: one set of process-based models (CM4 and ESM4) with the highest physical fidelity for ocean-driven hazards and hydrological dynamics. When CMIP6 ensemble tails produce physically implausible extreme projections, GFDL is the process-credibility filter. When CMIP6 ensemble mean underestimates compound event probability, GFDL's explicit co-occurrence physics provides the correction. The two models are not alternatives — GFDL is the quality control layer for CMIP6 ensemble tails.

CMIP6 Core Ensemble NOAA GFDL Physical Risk Lens
Model architecture 40+ models; emphasis on ensemble spread for uncertainty quantification 2 process-based models (CM4 + ESM4); emphasis on physical process fidelity
SSP coverage Full SSP1-1.9 through SSP5-8.5 with large ensemble SSP2-4.5, SSP3-7.0, SSP5-8.5 only — compute budget limits low-forcing runs
Compound events Compound events modelled probabilistically from ensemble statistics Co-occurrence explicitly modelled from coupled ocean-atmosphere-land dynamics
Best hazard types Temperature, precipitation, broad physical risk — all hazards at global scale Tropical cyclones, storm surge, hydrology, sea level rise, compound flooding — process-credible
Regulatory role IPCC AR6 and NGFS backbone — primary regulatory disclosure reference US DOE/FERC/FEMA expert elicitation anchor; insurance tail calibration reference
Use CMIP6 for scenario-conditioned long-run physical risk across the full SSP pathway space and for regulatory TCFD/NGFS alignment. Use GFDL as the process-credibility anchor for specific hazard types — tropical cyclones, compound flooding, sea level rise beyond 2050, and drought-heat co-occurrence — where physical process realism matters more than ensemble breadth. The combination is CE's standard for insurance tail calibration and long-duration infrastructure asset risk.

Methodology

The NOAA Geophysical Fluid Dynamics Laboratory (GFDL) produces process-based earth system models (CM4 and ESM4) with particular strength in hydrology, ocean heat content, and coupled atmosphere-land dynamics. GFDL models are considered the most physically credible for compound events involving multiple interacting systems — tropical cyclones, storm surge, riverine flooding, and drought — that cause the highest insurance and infrastructure losses. CE uses GFDL for physical plausibility validation: when CMIP6 ensemble outputs produce extreme tails, GFDL process fidelity provides a sanity check. GFDL hazard signals are generally more conservative than CMIP6 ensemble tails but more credible for hydrology-sensitive sectors.

Key Mechanisms

  1. Coupled ocean-atmosphere dynamics: GFDL explicitly models ocean heat uptake and release, creating more credible tropical cyclone intensity projections
  2. High-resolution hydrology: GFDL's land surface model produces river discharge and groundwater recharge projections at finer resolution than most CMIP6 models
  3. Sea level rise from ice dynamics: GFDL's ice sheet model provides physically grounded sea level rise projections beyond 2050
  4. Compound event simulation: the model captures co-occurring multi-hazard events (wind + storm surge + precipitation) that produce tail losses
  5. Process plausibility check: GFDL outputs are used to constrain the CMIP6 ensemble tail by eliminating physically implausible outliers
  6. Land-atmosphere teleconnections: soil moisture, vegetation cover, and large-scale atmospheric circulation interact — drought-heat wave co-occurrence probability is explicitly modelled, creating compound event amplification that single-hazard models miss
  7. Ocean heat content vertical mixing: deep ocean heat uptake rate determines the realised vs. committed warming differential — GFDL's ocean model captures the multi-decadal lag between emissions and surface temperature, affecting long-run sea level projections
  8. Urban and coastal morphology: high-resolution GFDL regional configurations capture how urban geometry, coastal shape, and land use modulate local hazard intensity — relevant for facility-level risk in coastal industrial zones

Score & Confidence Methodology

Hazard scores (0–1) are calibrated against IPCC AR6 WG2 Table 16.SM.1 industry-sector exposure bands. Transition pressure scores use NGFS Phase IV scenario families. Confidence intervals are asymmetric where IPCC likelihood language (likely/very likely) maps to the p17–p83 and p5–p95 ranges. Scores are not actuarially certified — see Known Limitations.

Known Failure Modes

  • Computationally expensive: fewer scenario runs than CMIP6 ensemble, creating less coverage of SSP pathway diversity
  • Less suited to rapid policy scenario iteration — designed for physical process analysis rather than flexible policy scenario exploration
  • Transition pressure signals are derived indirectly from implied carbon price pathways rather than from the model's physical outputs directly
  • Regional high-resolution configurations (CM4-HR) are computationally expensive — very few ensemble members exist, limiting uncertainty quantification to a narrower range than CMIP6 provides
  • Raw GFDL outputs require expert post-processing — not directly plug-and-play without specialist interpretation; less accessible than CMIP6 ensemble products for non-specialist users

Best For

hydrology, coupled earth-system dynamics, and process credibility

Strengths

  • Highest physical process credibility for hydrology, ocean-driven hazards, and compound events — particularly relevant for insurance tail calibration
  • Sea level rise and coastal flooding projections are the most rigorous available for long-duration asset risk (Prologis, Brookfield coastal portfolios)
  • Compound event modelling captures correlated multi-hazard scenarios that single-hazard models systematically underestimate
  • Best-in-class for US Gulf Coast and Caribbean hurricane basin hazards — CM4 validated against 40+ years of historical hurricane track and intensity data; the reference for Gulf Coast energy infrastructure and Florida coastal portfolio risk
  • CM4 and ESM4 reach 25 km grid resolution in regional configurations — the highest-resolution coupled earth system models in operational CE use, enabling facility-level coastal and hydrological hazard assessment
  • Accepted as expert elicitation anchor in US regulatory proceedings (DOE, FERC, FEMA climate risk guidance) — GFDL outputs carry direct regulatory credibility for US infrastructure asset disclosure

Maturity & Validation

Model era: Current • Status: active
Core models are internally cross-validated against institutional benchmarks. Advanced modules (DSGE, Monte Carlo, Catastrophe, Commodity) are prototype-grade — not yet independently peer-reviewed. View the full validation record at Validation Registry and current capability status at Capability Registry (JSON).

Scenario Coverage

IPCC SSP2-4.5 — core physical risk horizon for infrastructure and real estate investment IPCC SSP3-7.0 — compound hazard stress scenario; delayed action physical risk trajectory IPCC SSP5-8.5 — tail risk / extreme physical forcing scenario IPCC SSP1-1.9 / SSP1-2.6 — insufficient compute budget for full ensemble at low forcing NGFS Net Zero 2050 (as a transition scenario) — transition dynamics are outside GFDL's scope Near-term (1–5yr) physical risk — use ERA5-calibrated for observationally grounded near-term accuracy

GFDL is designed for physical process depth, not scenario breadth. For full SSP pathway coverage, use CMIP6 ensemble. GFDL is most valuable as the process-credibility anchor for CMIP6 ensemble tail validation.

Calibration Benchmarks

NOAA GFDL Technical Memoranda (CM4, ESM4 documentation) Core model validation: ocean heat content, tropical cyclone intensity, hydrological cycle, and sea level rise component verification
IPCC AR6 WG1 Chapter 9 (Ocean and Cryosphere) Sea level rise and ocean component credibility assessment; GFDL explicitly cited for ice dynamics and deep ocean mixing
PCMDI CMIP6 Model Evaluation Metrics GFDL model performance ranking within CMIP6 ensemble — basis for weighting GFDL relative to ensemble mean
Swiss Re Sigma Compound Event Loss Database Compound event probability calibration: GFDL co-occurrence statistics validated against 1,200+ historical compound event loss records
FEMA NFIP Flood Insurance Claims (coastal zone) Coastal flooding and storm surge calibration: GFDL coastal inundation projections validated against historical NFIP claims patterns
Industry Signal Dashboard — projected signals from this model across all tracked industries
Physical Hazard Pressure by Industry
Physical hazard index (0–1) indicating asset and operational exposure to climate-related physical risks.
Transition Pressure by Industry
Regulatory and market pressure from the low-carbon transition — 0 (low) to 1 (high).
Adaptive Resilience by Industry
Resilience index (0–1) — the industry's estimated capacity to adapt to physical and transition risk.
Industry Context
Energy
GFDL's coupled ocean-atmosphere dynamics provide the highest-fidelity projections of tropical storm intensification (offshore rig exposure for Aramco and ExxonMobil) and sea surface temperature trends (hurricane tracks affecting Gulf Coast energy infrastructure). GFDL's coastal flooding and storm surge projections are the most credible for NextEra's Florida coastal solar and wind assets.
Agriculture
GFDL provides the most physically credible projections of hydrological change for agriculture — river discharge, groundwater recharge, and drought intensity for irrigation-dependent regions. These projections are critical for JBS's Brazilian cattle operations, Cargill's corn belt sourcing, and Bunge's South American grain origins. GFDL's agricultural hazard signal is typically the most extreme of the three CE climate models for water-stressed regions.
Manufacturing
GFDL's physical process credibility benefits industrial facilities requiring water-intensive cooling or processing. Its superior land surface model captures water availability risk for BASF's Rhine-dependent Verbund system and Rio Tinto's water-stressed Pilbara operations. GFDL's soil erosion and geotechnical risk projections are also relevant for ArcelorMittal's mine-to-port infrastructure.
Transport
GFDL provides the most credible projections of tropical cyclone track and intensity changes affecting shipping routes (Maersk's trans-Pacific and Atlantic corridors) and aviation disruption (Delta's Gulf Coast hub exposure). Its sea level rise projections for port infrastructure (Maersk terminal network) are the reference for long-duration asset risk. GFDL's transport hazard signals reflect physical reality of increased extreme weather on route scheduling.
Insurance
GFDL produces the most process-grounded probability distributions for extreme weather events that drive insurance loss scenarios. Its coupled model architecture captures co-occurring multi-hazard events (compound flooding + wind) that cause the highest insurance losses, making it the tail-risk calibration anchor for Allianz, Munich Re, and Swiss Re's catastrophe models in CE's insurance sector calibration.
Real Estate
GFDL's coastal flooding and sea level rise projections are the most physically grounded of the three CE climate models. For Prologis's logistics hubs, Brookfield's diversified global portfolio, and British Land's London flood risk, GFDL provides the highest-confidence assessment of chronic inundation risk for long-duration real estate holdings. Its urban pluvial flood hydrology also calibrates Vonovia's German residential flood exposure.
 Back to Model Catalog  Open in Workbench