What is statistical downscaling?
In short: Statistical downscaling converts coarse General Circulation Model output into climate information for a finer location or station. It calibrates a statistical relationship between historical model data and local observations, then applies that relationship to model projections. The result is more suitable for local impact studies, but it still depends on the quality of the observations and model simulations.
How statistical downscaling works
A typical workflow aligns observed station records with historical GCM data for a common calibration period. A transfer function is estimated from those paired data, evaluated on independent historical data when possible, and then applied to a future CMIP5 RCP or CMIP6 SSP simulation.
What it can and cannot do
Statistical downscaling is computationally efficient and can produce station-scale time series for many models and scenarios. It does not repair every physical error in a GCM, and its future use assumes that the calibrated statistical relationship remains useful under a changing climate.