3.1 Flow over a bell-shaped hill

**Figure 2:** Simulation results of the Boulder windstorm. The topography is represented by a bell shaped hill with a height of 2000 m. Initial conditions for temperature and wind speeds were taken from a radio sounding at Grand Junction. The figures shows the values after 10 hours of simulation.
$\begin{figure} \begin{center} \leavevmode \epsfysize=20cm \epsffile{boulder-fig.eps} \end{center} \end{figure}$

**Figure 3:** Reconstruction of potential air temperature (upper graph) and horizontal wind (lower graph) speed on 11 January 1972 from aircraft measurements, according to Lilly (1978). Dotted lines represent the aircraft flight path.
$\begin{figure} \begin{center} \leavevmode \epsfysize=19cm \epsffile{boulder-mess.eps} \end{center} \end{figure}$

In January 1972, Boulder (Colorado), located on the eastern slope of the Rocky mountains experienced a severe Chinook windstorm with wind speeds up to 60 ms ^{- 1} which is unique in the sense that many experimental data from that event are available (Lilly and Kennedy; 1973; Lilly and Zisper; 1972; Lilly; 1978). The potential temperatures and horizontal wind speeds, found by aircraft measurements are represented in Figure 3. The general situation can be approximated with a bell-shaped hill with a height of 2000 m and by initializing the model with the wind speeds and temperatures measured by radio soundings. This case has been used regularly to test the proper behavior of mesoscale models (e.g. Buty; 1988; Thunis; 1995), and therefore has been applied in the validation process of the dynamical part of Metphomod , too. Figure 2 shows fields of virtual potential temperature, wind speeds in W-E direction, and vertical wind speeds, respectively. The main features that were found in the campaign are well reproduced: A strong wind acceleration behind the ridge with a wind maximum at the ground of about 55 ms ^{- 1}; strong negative vertical wind speeds of about - 8 ms ^{- 1} after the ridge, followed by even higher positive vertical wind speeds; the generation of gravity waves. The gravity waves, however, seem to be damped down too quickly. This may be due to the advection scheme which in its present implementation in Metphomod applies Smolarkievicz's (1984) anti-diffusion correction only in the horizontal directions and therefore has important numerical diffusion in the z-direction in a situation with high vertical wind speeds.