xclim

xclim, developed by Ouranos (Québec's climate consortium), computes a large library of climate indicators — heat waves, growing degree days, precipitation extremes, dry spells — on xarray datasets with proper units handling, metadata conventions and dask-powered scaling to large ensembles. Its sdba module provides the standard bias-adjustment methods (quantile mapping and successors) needed before projections touch engineering models.

It is the closest thing to an operational standard for climate-projection post-processing in Python: health-checked inputs, CF-compliant outputs, and the numerical care this niche often lacks.