The Asian Development Bank's Risk Explorer gives its staff and partners a common view of the climate hazards its investments face. FutureWater produces three global hazard datasets for the platform, covering extreme heat, drought and wildfire. Each is derived from bias-corrected climate model ensembles and delivered for a historical baseline and four Shared Socioeconomic Pathways across seven overlapping thirty-year horizons to the end of the century. The products are hazard layers rather than risk layers: they describe the intensity and frequency of the hazard at a location, and deliberately contain no population, asset or vulnerability information. Each dataset comes with a methodology report documenting how every layer was derived, the validation behind it, its known limitations, and what is needed to reproduce it.
Development banks need a consistent way to screen the climate hazards their investments face, across every country they lend in and on a common basis. The Asian Development Bank’s Risk Explorer serves that purpose, and it depends on hazard datasets that are global, current, and traceable to a documented method.
FutureWater produces three of those datasets: extreme heat, drought and wildfire. Each is delivered for a historical baseline and for four Shared Socioeconomic Pathways, across seven decade-centred thirty-year horizons running to the end of the century. Because consecutive horizons overlap by twenty years, they are documented as a rolling view rather than as independent samples.
The heat product is built from the NASA NEX-GDDP-CMIP6 ensemble and reported on three thermal lenses computed in parallel from the same daily inputs: dry-bulb maximum temperature, a simplified wet-bulb globe temperature, and wet-bulb temperature. They are not alternatives to choose between. Each answers a different question about heat, and the difference between them carries information about humidity. Each lens in turn carries four index families, separating how extreme the rare event is, how often a damaging threshold is crossed, and how far a heatwave departs from what a location is used to.
The drought product uses the Standardized Precipitation-Evapotranspiration Index to characterise drought frequency, duration, severity and intensity, with events detected and a climatology built before bias correction against an observational baseline. The wildfire product derives bias-corrected fire weather from the NASA NEX-GDDP-FWI archive and layers it in three tiers: an absolute hazard, a biome-relative hazard that asks what is unusual for a given ecological zone, and a future-fuel-aware hazard that accounts for changing vegetation.
Common to all three is an emphasis on traceability. Each dataset is bias-corrected against an observational baseline, checked through a structured set of validation strands, and shipped with documented metadata conventions, a file manifest and checksums so that a delivery can be verified. The methodology reports record the choices that were considered and rejected alongside those adopted, so a reviewer can follow the reasoning behind each fork.
