D2: Inventory of downscaling methods, validation studies and gaps of knowledge

The inventory of downscaling methods will encompass the range of state-of-the-art RCMs with a focus on European models, but include also non-European models; traditional PP statistical downscaling methods such as linear and non-linear regression (e.g., artificial neural networks), weather typing and analogue methods; MOS methods such as variants of the direct method (including scaling) and quantile mapping methods. Additionally, the inventory will search for new methods based on stochastic models and downscaling of full PDFs, and methods for downscaling daily time series to sub-daily time series. The inventory will group the methods according to the downscaling approach as well as to its applicability (e.g., number of downscaled variables, local vs. spatially distributed, time scales provided). As this inventory is concerned with methods for different purposes, it will be carried out by WG3-5, coordinated by WG1. WG3 is responsible for temporal and spatial variability, WG4 for extremes, and WG5 for sub-daily scales.