Added value of Statistical Downscaling for seasonal forecasting in Argentina

BALMACEDA HUARTE, R; BETTOLLI MARIA LAURA; PENALBA, O.C.

La Paz

Workshop; CORDEX Central America and South America Training Workshop on Downscaling Techniques; 2018

CORDEX-WCRP

Global Climate Models (GCMs) are the primary tools used to generate weather and climate predictions at different forecast horizons, from intra-seasonal to centennial scales. However, it is well recognized that these models are unable to provide information at the spatial scale required by many stakeholders. Hence, downscaling methods are necessary for adapting the global model predictions to smaller spatial scales, providing suitable products for a range of applications. The objective of this work was to explore the added value of multiple linear regression as a statistical downscaling (ESD) tool for daily maximum and minimum temperatures using seasonal hindcasts from different global circulation models. To this end, three meteorological stations located in the central region of Argentina (Mendoza Observatory, Santa Rosa Aero and Central Observatory of Buenos Aires) were used. Also daily fields of geopotential height at 500 hPa, air temperature and specific humidity at 850 hPa, and pressure at sea level from Reanalysis 1 (NCEP) were used as large-scale predictors. The period analyzed corresponds to the years 1978-2015 in the austral summer (December-January-February). The statistical model was calibrated in the period 1978-2000 based on the stepwise regression method and the cross-validation technique. Additionally, the 2001-2016 period was used as an independent period for validation.