Prospects for downscaling seasonal precipitation variability using conditioned weather generator parameters

This paper explores the use of synoptic-scale predictor variables to downscale both high- and low-frequency components of daily precipitation at sites across the British Isles. Part I investigates seasonal and inter-annual variations in three weather generator parameters with respect to concurrent variations in a North Atlantic Oscillation (NAO) index and area-average sea surface temperature (SST) anomalies. Marked spatial gradients were found in the strength of the associated correlation fields using rainfall data for the period 1961–90. For example, the persistence of winter wet-spells was most strongly correlated with the NAO index in NW Scotland, and the persistence of autumn dry-spells with SST anomalies in SE England. At such locations, North Atlantic conditioning accounted for over 40% of the inter-annual variability of precipitation occurrence. In Part II, three downscaling models were compared using independent daily precipitation data for sites located in the regions of strongest North Atlantic forcing. The parameters of Model M were implicitly conditioned by three regional airflow indices; the parameters of Model X were explicitly conditioned by either the NAO index or SST anomalies and daily vorticity; and the parameters of Model U (a three-parameter stochastic rainfall model) were unconditional. Overall, the conditional models displayed greater skill for monthly rainfall statistics relative to Model U (the control), but still did not completely remove overdispersion. On comparing Models M and X, it was evident that explicit conditioning did bestow additional advantages for the chosen sites and seasons of greatest forcing. However, further research is required to determine the generality of these results for other regions and periods of the rainfall record.