Statistical downscaling of daily precipitation using daily airflow and seasonal teleconnection indices

Monthly or seasonal climate variability is seldom captured adequately by high-resolution statistical downscaling models. However, such deficiences may, in fact, be an artefact of the failure of many downscaling models to incorporate appropriate low-frequency predictor variables. The present study explores the possibility of using variables that characterise both the high- and low-frequency variability of daily precipitation at selected sites in the British Isles. Accordingly, 3 statistical downscaling models were calibrated by regressing daily precipitation data for sites at Durham and Kempsford, UK, against regional climate predictors for the period 1881-1935. Model 1 employed only 1 predictor, the daily vorticity obtained from daily grid-point mean-sea-level pressure over the British Isles. Model 2 employed both daily vorticity and seasonal North Atlantic Oscillation Indices (NAOI) as predictors. Finally, Model 3 employed daily vorticity and seasonal North Atlantic sea-surface temperature (SST) anomalies as predictors. All 3 models were validated using daily and monthly precipitation statistics at the same stations for the period 1936-1990. Although Models 2 and 3 did yield improvements in the downscaling of the monthly precipitation diagnostics, the enhancement was only modest relative to Model 1 (the vorticity-only model). Nonetheless, the preliminary results suggest that there may be some merit in using North Atlantic SST series as a downscaling predictor variable for daily/monthly precipitation in the UK. However, further research is required to determine whether or not the inclusion of teleconnection indices in downscaling schemes leads to better representations of low-frequency variability in both present and future climates when General Circulation Model outputs are employed.