Methodology
The Intelligence Layer Between
Climate Science and Financial Decisions.
ClimRisk translates peer-reviewed climate science into asset-level financial risk, expressed in dollars, not degrees. This page is the whitepaper version: the pipeline, the data behind it, the financial math, and the frameworks it was built to satisfy, for the risk managers and quants who need to defend the number before they sign off on it.
How It Works
From climate science to financial decisions.
Seven steps, run in sequence, every time. Steps three and four are where most of the engineering effort has gone: turning a hazard probability into a number that shows up on an income statement is the hard part, not the hazard data itself.
Climate Hazard Quantification
Site-level flood depths (JRC river, Deltares coastal with sea-level rise), cyclone wind from NOAA IBTrACS tracks since 1980, WRI Aqueduct 4.0 water risk, plus 25 parametric hazard functions scaled by IPCC AR6 warming and CMIP6 projections.
JRC · Deltares · IBTrACS · CMIP6Asset-Level Exposure Mapping
Each asset's coordinates are read against the hazard maps, local flood defences (FLOPROS), 10 m terrain, Sentinel-1 radar and Sentinel-2 imagery, and ESA WorldCover land cover.
GIS · FLOPROS · Sentinel · WorldCoverOperational Disruption Modelling
ProprietaryAn equipment sensitivity matrix combined with sector-specific vulnerability curves calculates business interruption probability, asset by asset.
BI Model · Sensitivity MatrixRevenue & Cash Flow Translation
ProprietaryAnnual loss fractions are mapped to revenue impact and validated against IPCC AR6 observed regional losses, not modelled in isolation.
DCF · ALF · Revenue ModelEnterprise Value Impact
Climate-adjusted EV, Value at Risk, and stressed NPV are computed across three time horizons: 2030, 2040, and 2050.
EV · VaR · NPVAdaptation ROI Analysis
A cost-benefit read on adaptation measures, flood walls, cooling systems, supply chain diversification, each with its own payback period.
Adaptation · ROI · PaybackAudit-Ready Report Output
IFRS S2, TCFD, CSRD and SEBI BRSR-aligned reports with named sources, Monte Carlo ranges on loss figures, and the model's caveats stated alongside.
IFRS S2 · TCFD · CSRDData Provenance
Transparent by design.
Every number the engine produces is traceable to one of these sources. Risk managers need to know exactly which models sit underneath a disclosure so they can defend it to their own auditors, so here they are, grouped by what they actually do.
Physical Climate Models
CMIP6 HighResMIP
Coordinate-level temperature and precipitation change (MRI-AGCM3-2-S with fallbacks) via the Open-Meteo climate API.
NOAA IBTrACS · JRC · Deltares
Cyclone tracks since 1980; river flood depth maps (10–500-year); coastal flood depths 2018 and 2050 with sea-level rise.
Earth Observation
Copernicus Sentinel-1 / Sentinel-2 · ESA WorldCover
10 m radar flood history, optical imagery and land cover at each site. NASA POWER and FIRMS for weather and active fires.
WRI Aqueduct 4.0
13 water-risk indicators for 68,506 basin × admin-1 units, with projections to 2030, 2050 and 2080.
Financial Transition Scenarios
NGFS Phase 5
7 scenario pathways from 3 models (GCAM, MESSAGEix-GLOBIOM, REMIND-MAgPIE), with carbon prices, sector activity and grid intensity.
World Bank Carbon Pricing Dashboard
94 carbon taxes and trading systems, matched to each asset's jurisdiction and sector.
Carbon Accounting Factors
UK DESNZ GHG Conversion Factors 2026
Fuels, vehicles, refrigerants, travel, freight, waste, water, materials and well-to-tank — CH4 and N2O restated to IPCC AR6.
US EPA eGRID2023 · Green-e 2025 · AIB 2025
Grid factors by eGRID subregion and European country; US and European residual mixes for market-based Scope 2.
US EPA USEEIO v1.3
Spend-based supply-chain factors for 1,016 industries (2022 USD), deflated with US CPI.
The Engine
CRI Engine
Tested against measured cyclone winds, FEMA flood claims and EPA-reported emissions — results, including the weak ones, on the Validation page.
Product Carbon Footprint (LCA)
ISO 14067-aligned cradle-to-gate footprints with IPCC 2006, GCCA, Ember and DESNZ 2025 factors, uncertainty bands and an EU CBAM view.
ML & Predictive Layer
CRI Trajectory Model
Gradient-boosted surrogate trained on 50,000 simulated company-years calibrated to NGFS pathways and EM-DAT losses, forecasting the CRI score to 2050 with confidence bands.
ClimateBERT Disclosure Scanner
Four Hugging Face ClimateBERT classifiers score a company's own disclosures for climate relevance, sentiment, and net-zero commitment specificity.
Emissions Estimator
Trained on EPA GHGRP observed emissions matched to company revenue; used only where reported or facility-level emissions are missing, and labelled as an estimate.
The Financial Math
DCF, VaR, and NPV, stated plainly.
We don't publish the proprietary scoring model itself, but the financial machinery around it is not a secret. This is what an auditor would need to know to follow the math.
Discounted Cash Flow
A full DCF runs the 2026 to 2050 horizon with a Gordon Growth terminal value. The discount rate is each company's own base WACC, typically 8 to 9 percent, stressed with a scenario premium and an asset-specific exposure premium on top.
Value at Risk
Portfolio VaR is computed at 95% and 99% confidence, weighted by exposure across the full asset book, and reported alongside the WACC uplift that's actually driving it.
Net Present Value
Stressed NPV is reported at three checkpoints, 2030, 2040, and 2050, so a board can see where in the horizon the exposure actually bites, not just the endpoint number.
The Output
Science this rigorous exists to satisfy frameworks this specific.
Everything above feeds the same disclosure engine. See the full breakdown of what each framework requires and what the engine hands back on the frameworks page.