The *CLIMTAG Precipitation* indicator represents the precipitation derived from ERA5 reanalysis (past) and CMIP6 climate projections (future) under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios.Annual map (ANNUAL_MAP): annual precipitation sum (mm/year) as spatial GeoTIFFs at reference years 1996â2086 (every 10 years), with ensemble percentiles P10/P25/median/P75/P90. ## Past (ERA5 reference)Daily precipitation totals are fetched from the ERA5 reanalysis (CDS) for the reference temporal extent (1981â2010) window covers the historical anchor). The daily timeseries is temporally aggregated to annual or dekadal sums. A climatology is then computed using a sliding window centred on each reference year. The ERA5 historical reference climatology is downscaled to ~30 arc-second resolution using the GMTED2010 orography dataset. Spatially varying sensitivity parameters for orography (β_oro) and distance from the coastline (β_coast) are estimated from a ~180 km rolling-window ordinary least-squares regression applied to the coarse ERA5 field, and are used to correct for sub-grid elevation and coastal gradients not resolved at the coarse scale. ## Future (CMIP6 projections)For each CMIP6 model/member, the full projection timeseries is fetched and bias-corrected against ERA5 using empirical quantile delta mapping: this is done for each future timestep according its percentile within a sliding-window cumulative distribution function of the future biased timeseries (30-year window, 5-year step), and the correction is applied according to this percentile rank. Therefore, the bias for each percentile rank is calculated from difference between model historical quantiles and the ERA5 quantiles (after upscaling to the model grid) over the same reference period (1981â2010). The bias correction procedure is done separately for each month of the year using 3-month pooling windows to stabilise tail estimation. The corrected timeseries are then temporally aggregated and a climatology is computed over the same 30-year sliding window. Results are warped to a common 1° grid before ensemble statistics (P10, P25, median, P75, P90) are computed across all model members. The final projected climatologies are produced by a delta-mapping approach that combines the coarse-scale ensemble signal with the downscaled ERA5 field.