Assessment of uncertainties in simulated European precipitation

<p>The research presented in this thesis is aimed to understanding the changes and the simulation of precipitation in Europe. A correct representation of simulated (trends in) European precipitation is important to have confidence in projections of future changes therein. These projections are relevant for different hydrological applications. Among others, simulated changes of summer drying are often accompanied by an enhanced increase in air temperatures [<em>Zampieri et al.</em>, 2009]. This can be expected to have large impacts on society and ecosystems, affecting, for example, water resources, agriculture and fire risk [<em>Rowell</em>, 2009]. Projections of changes in extreme precipitation are critical for estimates of future discharge extremes of large river basins, and changes in frequency of major flooding events [e.g. <em>Kew et al.</em>, 2010].<br/> <br/> The subjects that are studied in this thesis are divided in three parts: (1) evaluation of 20th century European precipitation trends; (2) effect of general circulation model (GCM) spatial resolution on simulated western European winter precipitation in the current climate; and (3) effect of GCM spatial resolution on simulated future summer drying in central and southern Europe.<br/> <br/> In the first part of the thesis (chapters 2 and 3) an investigation of (extreme) precipitation trends in multi-model ensembles including both global and regional climate models is performed. The results show that these models fail to reproduce the observed trends over (parts of) the past century. In many regions the model spread does not cover the trend in the observations: the models significantly underestimate the observed trend. A misrepresentation of large scale atmospheric circulation changes in climate models is found to be responsible for the underestimation of winter precipitation trends in Europe over the past century. Additionally, the underestimation of trends in winter precipitation extremes in the Rhine basin is directly related to this as well. In summer a misrepresentation of sea surface temperature (SST) trends is responsible for the underestimation of summer precipitation trends along the coastal regions of western Europe.<br/> <br/> The second part (chapter 4) investigates the effect of GCM spatial resolution on modeled precipitation over Europe using an atmosphere-only GCM at two resolutions (EC-Earth, ~25 km and ~112 km horizontal resolution). The results show that the high resolution model gives a more accurate representation of northern and central European winter precipitation. The medium resolution model has a larger positive bias in precipitation in most of the northern half of Europe. Storm tracks are better simulated in the high resolution model, providing for a more accurate horizontal moisture transport and precipitation. A decomposition of the precipitation difference between the medium- and high resolution model in a part related and a part unrelated to a difference in the distribution of vertical atmospheric velocity confirms that the reduced precipitation in the high resolution model is likely the result of a reduced moisture transport at this resolution: the precipitation difference in this area in unrelated to a difference in the distribution of vertical atmospheric velocity. In areas with orography the change in vertical velocity distribution is more important.<br/> <br/> Using the same atmosphere-only model, the third part (chapter 5) of this thesis investigates the influence of GCM spatial resolution on the simulated future summer drying of central Europe. High resolution models have a more realistic representation of circulation in the current climate and could provide more confidence on future projections of circulation forced drying. The results show that the high resolution model is characterized by a stronger drying in spring and summer, mainly forced by circulation changes. The initial spring drying intensifies the summer drying by a positive soil moisture feedback. The results are confirmed by finding analogs of the difference between the high and medium-resolution model circulation in the natural variability in another ensemble of climate model simulations. In current climate, these show the same precipitation difference pattern resulting from the summer circulation difference. In future climate the spring circulation plays a key role as well. It is concluded that the reduction of circulation biases due to increased resolution gives higher confidence in the strong drying trend projected for central Europe by the high-resolution version of the model.<br/> <br/>