A high level design principle of standardisation was applied in all methodologies, to provide data that is easily comparable across hazards, and over time. Available data is highly varied between hazards, making standardisation an important design principle that applies across all hazards.

Each hazard is represented by two complementary dimensions for assessing the potential financial implications of physical risk: a normalised intensity index, which captures maximum potential severity, and an annual exceedance probability, which captures how often damage-relevant conditions are expected to occur.

Rather than focusing on absolute hazard levels, the methodology emphasises changes relative to a historical baseline, isolating the climate-driven signal from local infrastructure, land use, and defence features that global models cannot reliably resolve.

To preserve meaningful differentiation across climate pathways, hazard intensities are normalised using scenario-consistent global reference maxima, rather than local or historical scaling. This ensures that higher-emissions pathways register proportionally higher hazard levels and enables robust comparison across scenarios, supporting consistent interpretation of hazard escalation over time.

To model global Wildfire risk, Emmi builds on the Fire Weather Index (FWI) from CMIP6 climate projections (Quilcaille et al., 2023), which provides daily meteorological fire danger at ~0.25° resolution spanning 1850-2100 for RCP 2.6, 4.5, 6.0, and 8.5 scenarios. The FWI integrates temperature, relative humidity, wind speed, and precipitation through the Canadian Forest Fire Weather Index System which is the global standard for fire danger assessment adopted by 40+ countries (van Wagner, 1987; Di Giuseppe et al., 2016).

Meteorological fire danger alone cannot predict actual fire occurrence. Emmi grounds FWI-based projections through calibration to 24 years of satellite-observed burned area from MODIS (2001-2020) and Sentinel-3 (2021-2024), capturing major fire events, and by adding vegetation and elevation fuel modifiers. These calibrations establishes location-specific relationships between meteorological conditions and actual burning, integrating unmeasured factors including fuel availability, human ignition sources, and suppression effectiveness.

To model global Tropical Cyclone risk, Emmi builds on the STORM v3 synthetic tropical cyclone database, which provides 10,000 years of statistically robust storm tracks across all six global basins, far exceeding the ~40 years of quality observational data (Bloemendaal et al., 2020).

These synthetic tracks are combined with basin-specific climate scaling factors from Knutson et al. (2020) that reflect CMIP6 projections for intensity increases, frequency changes, and poleward migration under RCP 2.6, 4.5, 6.0, and 8.5 scenarios.

To model global Flooding (Fluvial and Coastal) risk, Emmi builds on the WRI (World Resource Institute) Aqueduct Floods v2 dataset, which provides globally consistent coastal and fluvial flood depths at ~1 km resolution for nine standard return periods (2–1000 years), across a historical baseline and future time slices (2030, 2050, 2080) under RCP4.5 and RCP8.5 (Ward et al., 2020; Winsemius et al., 2016).

The fluvial component is based on the GLOFRIS framework, combining the PCR-GLOBWB 2 global hydrological model with CaMa-Flood inundation routing (Winsemius et al., 2013; Sutanudjaja et al., 2018).

The coastal component builds on the Global Tide and Surge Reanalysis (GTSR) and superimposes sea-level rise projections consistent with CMIP5 scenarios (Muis et al., 2016; Jevrejeva et al., 2018).