Design temperature — CHELSA bio5 (current)
How hot the afternoons get in a normal hot month — the heat a site's cooling and its workforce have to cope with.
Source: CHELSA v2.1 bio5, ensemble mean of GFDL-ESM4 / MPI-ESM1-2-HR / MRI-ESM2-0
Year: 2011-2040 climatology
Category: Physical risks · Direct Physical Loss engine inputs
Coverage: Global
Format: Raster grid (~2 km)
Used in risk analysis: Yes — gates Mitigating services (Local micro and meso climate regulation) at ≥ 33 °C
What it shows
How hot the afternoons get in a normal hot month, in °C: the average daily maximum temperature of the warmest month of the year. It is a design temperature — the level cooling equipment is sized for and a workforce has to work through — not a record ever measured. Because it averages the days of that month, it sits below its hottest single afternoon.
How it is built
CHELSA v2.1 downscales global climate-model output to ~2 km. Darwin averages three models and reads the current-conditions band, the 2011-2040 climatology whose midpoint (2025) is closest to the reference year; the emissions scenarios have not meaningfully diverged that early.
How to read it
Read it as a level: the value is the temperature itself, in °C.
Class thresholds
A site is flagged on Mitigating services when the design temperature reaches 33 °C.
How this threshold was calibrated
It replaces a ≥ 35 °C rule that was read on an end-of-century projection, and it is set to flag the same share of land as that rule did — so moving the flag onto today's climate neither widens nor narrows it. This is a Darwin calibration; 35 °C remains the literature-facing heat-stress figure for anyone who prefers the published anchor to the equal-share choice.
The measurement behind it, for audit: the old rule covers 4.04 % of land pixels, and the current climatology reaches that same share at 32.79 °C. Keeping 35 °C on this band would have shrunk the flag by a third, to 2.69 %.
Why this layer
It is the raster the Direct Physical Loss engine scores heat from, and it describes the present — which is what a screening view claims to show. See Max temperature of warmest month, SSP3-7.0 2071-2100 for the end-of-century projection, which stays contextual.