D3.1.d: Regional seasonal predictions, with uncertainty estimations, in the target regions, and inclusion in the QWeCI Statistical Downscaling Portal, accessible to partners in user-friendly formats

Regionalized seasonal predictions of maximum temperature and precipitation for a set of selected stations within the QWeCI countries were obtained using state-of-the art statistical downscaling methods (analogs, regression, and GLMs). First, the downscaling methods were calibrated under the Perfect Prognosis approach, that is, ERA-Interim reanalysis was used as predictor data, identifying the optimum predictors for each variable in each region. Then, the resulting statistical methods were applied to the outputs of the ENSEMBLES Stream 2 multi-model experiment. The skill of the Stream 2 raw and downscaled predictions was evaluated and compared against the Perfect Prog. results, considered as the limit of attainable skill. Uncertainties associated with the different models, downscaling methods and predictor datasets were also assessed. Finally, the experiments here presented are available through the QWeCI downscaling portal, where users can downscale the regional projections or even try alternative downscaling experiments in an interactive way.