The Spatial Distribution of Precipitation in Scania, Southern Sweden. Observations, model simulations and statistical downscaling

Abstract: This thesis focusses on the spatial distribution of precipitation within Scania and on its relation with large-scale climate. The spatial distribution of Scanian precipitation was analysed using a dense rain-gauge network of daily measurements. The performance of the Rossby Centre regional Atmospheric model in simulating the observed spatial distribution was evaluated. The influence of the regional atmospheric circulation on the observed precipitation distribution was assessed and statistical downscaling models, based on monthly large-scale climate, were established for mean precipitation and frequency of wet-days. The models were used to construct future estimates of these statistics, based on a GCM projection. It was found that the spatial co-variability of the monthly precipitation within Scania is high. Topography is the major factor influencing the precipitation distribution. Furthermore, there is a significant difference between the southeastern and the northwestern part of Scania related to the land/sea distribution in the area. Consequently, the annual mean precipitation in Scania has a spatial variation with a typical length scale of 20-35 km elongated in the N/NW to S/SE direction. The spatial patterns are not well captured by the RCA1 dynamical downscaling model. This may partly be caused by a too coarse resolution (44 km) of the RCA1. In total, 80-90% of the observed monthly meso-scale precipitation variation of Scania can successfully be related to, and explained by, the regional atmospheric circulation indices. This indicates that the regional atmospheric circulation is an important predictor in the statistical downscaling of precipitation. The daily precipitation patterns are also largely influenced by the pressure patterns in regional scale. Daily mean precipitation and the frequency of wet-days could be successfully modelled by the statistical downscaling procedure, using the monthly predictors of large-scale precipitation, relative humidity and atmospheric circulation indices. The downscaled scenario shows a significant increase of the annual daily mean precipitation by about 10 % and a decrease in frequency of wet-days by 1 % between the control (1961-90) climate and the scenario (2041-70) climate. Both daily mean precipitation and frequency of wet-days increase during winter and decrease during summer. The results also indicate an increase in the intensity of the precipitation, especially during winter. A winter increase and summer decrease of the westerly flow seems to be the major cause of the changes.