Mountain–valley precipitation differences in the northern Alps: an exemplary high-resolution modeling study

Abstract

This paper investigates the dependence on environmental conditions of altitudinal precipitation differences in the northern Alps, based on high-resolution numerical simulations with the MM5 model for a selected region in the Bavarian Alps (Zugspitze mountain and surrounding valley stations). Three exemplary precipitation events representing climatological regimes with different orographic enhancement characteristics are selected. After validating the MM5 precipitation fields against the available surface observations, the model results are used to analyse the interactions of atmospheric dynamics and cloud microphysics with the local orography. The first two cases (19–22 March 1997, 05–09 February 1999) are characterized by a strong northwesterly or northerly flow, associated with large precipitation differences between the mountain and the surrounding valley stations. For these cases, the model results indicate a dominance of the classical seeder–feeder mechanism, with strong orographic lifting generating dense orographic clouds over each individual mountain ridge, which in turn intensify precipitation. The related surface precipitation maxima can be found near the mountain peaks or somewhat in the lee due to hydrometeor drifting. The third case (05–07 December 1992) represents conditions with relatively small (i.e. below climatological average) precipitation differences between the Zugspitze and the surrounding valley stations. For this event, the model results indicate that relatively weak ambient winds at and below Alpine crest level (700 hPa) were primarily responsible for the lack of substantial precipitation enhancement. Precipitation was nevertheless moderately intense because of strong frontal lifting at higher levels. In all three cases, the agreement between simulated and observed precipitation patterns is so high that there is good reason to expect that mountain–valley precipitation differences will be quantitatively predictable for nonconvective events once a sufficiently high model resolution is computationally affordable.