Gard@gfi.uib.no

1. Introduction and generaL CONDITIONS

Vertical gradients of air temperature and wind has strong influence on out-door sound pressure levels (Embleton (1996)).

In the study presented here, the mesoscale model MM5 (Grell et al., 1994) has been used to investigate its ability to reproduce surface inversion conditions. To test the model and to compare with measured data, a situation from 12 UTC 20 September to 18 UTC 21 September 1994 was chosen. The geographical area of interest has been Finnskogen in Hedmark County, NE of Oslo, Norway. In this locality, the ground is undulating and mostly covered with forest, rising gradually from the river Glomma to the Swedish border.

A high-pressure system was situated over southern parts of Norway during the period studied (Figure 1), and the wind speeds were low or moderate most of the time. Extensive measurements with a Tethersonde were carried out from 6 UTC to 16 UTC on 21 September. A strong inversion in the lowest 300-500 meters above ground was observed. According to classical theory (e.g. Whiteman (1982) or Stull (1988)), the strongest inversions normally appear just after sunrise. A simulated inversion in MM5 has been compared to the observed soundings. Further on, results from MM5 are used as input to a sound propagation model, OASES (Schmidt, 1999), to investigate the meteorological influence on the sound propagation.

Some changes are introduced in the model to more accurately describe the short wave radiation (SW) at sloping surfaces. When the slopes of the topography are accounted for in the atmospheric computations, SW patterns are clearly changed. Preliminary results of SW are presented here.

2. Model configuration

MM5 was set up with an outer mesh with 13.5 km horizontal resolution and an inner mesh with 500 meter resolution. The data was nested from ECMWF (European Centre for Medium-Range Weather Forecasts) data in four steps down to 500 meter grid distance (Figure 2). The number of grid points were 40 ´ 40 for all domains and 31 vertical levels (6 below 100 meters). The options of the parameterizations of sub-grid and physical processes, were in addition to the first order turbulence closure (Hong & Pan, 1982), an advanced 5 layer surface model (LSM) (see Chen and Dudhia (2001a, 2001b)) with prognostic equations for soil moisture and temperature. Explicit moist "physics" including ice phase was used. Cumulus parameterization (Grell et al. (1991)) has been used for the outer mesh (D = 13.5 km).

3. Results

Figure 3 shows the time-evolution of the inversion at Finnskogen 21 September 1994. The figure reveals a classical example of morning inversion break-up, as recorded by the Tethersonde. It is clearly visible how the ground is heated by solar radiation and how statically unstable air close to ground penetrates deeper and deeper into the inversion and destroys it from below.

The inversion seems to be fairly well simulated at 6 UTC and 7 and after 11 UTC. On the other hand, the destruction rate of the inversion between 8 and 10 UTC is not satisfactorily simulated. The inversion in the model breaks up too fast. There are some possible explanations for this misfit which need to be investigated further. One might be lack of soil water freezing and thawing in the model (see e.g. Viterbo et al. (1999). This effect will slow down the break-up since the temperature at ground will have to cross the "heat-capacity barrier" (Viterbo et al. (1999). Other effects could be connected to the turbulence parameterizations. In this simulation, the formulations used are based on Hong & Pan (1982). Studies have shown that the strength of the inversion is sensitive to the formulation of the fluxes of momentum and heat profiles (see Sorteberg (2001)). During strong stable stratification, the turbulent diffusion coefficients in the boundary layer used in the MRF-scheme (Hong & Pan, 1982) becomes to low. This could be improved by using other formulations, e.g. those applied in the operational ECMWF-model (Viterbo et al. (1999)). These heat-flux formulations are based on the dependency on Richardson number (Viterbo et al. (1999)) rather than the height above ground and the PBL height.

At 16 UTC the observed profile is neutral or close to neutral. In the model, however, the formation of a new surface inversion has already started. The response between short wave radiation and temperature fluctuations at ground seems to be too fast in the model. This could be due to the fact that the storage term for energy (see eq. (2), last term) becomes too small.

4 Effects of meteorology on sound Propagation

Sound propagation in the atmosphere is influenced by topography, ground conditions (nature of the ground type, e.g. snow, vegetation or asphalt) and weather conditions (refraction, diffraction and turbulent scattering in the atmosphere) (Embleton, 1996). Here we concentrate on the effect of atmospheric refraction, which has most influence on sound pressure levels. The Directional Sound Velocity profile, DSV, can be written as:

(1)


where T(z) is the air temperature profile in K, T₀ = 273,15  K, c₀ = 331,5 ms^-1 is speed of sound at T₀, V(z) is the wind speed profile, a (z) is the wind direction profile and f is the source to receiver bearing. Equation 1 shows that the DSV is more dependent on changes in wind speed and direction, than it is on air temperature. It is normally assumed that the DSV profile is range-independent, i.e., that V, a and T are horizontally homogeneous over the area considered, and thus only dependent on z (Hole & Mohr, 1999). 3D sound propagation models that take all relevant effects into account are yet to be developed.

Here, 1D calculations in 8 directions were interpolated horizontally to achieve a quasi-2D result (depending on range and angle). DSV profiles were calculated from MM5 atmospheric profiles in each case and used as input in the well-known OASES acoustic model package (Schmidt, 1999). The model atmosphere had six layers below 100 m above ground, using the same height intervals as in MM5. A two-layer ground

Figure 4: A) Plane view of TL at 100 Hz as predicted by OASES with input based on observation at 6 UTC.	B) Plane view of TL at 100 Hz as predicted by OASES with input based on MM5 output at 6 UTC.
C) Plane view of TL at 100 Hz as predicted by OASES with input based on observation at 12 UTC.	D) Plane view of TL at 100 Hz as predicted by OASES with input based on MM5 output at 12 UTC.

model with acoustical properties corresponding to a typical Norwegian forest floor was included (Hole et al. (1998)).

Figure 4 clearly demonstrate the strong effect of the temperature inversion on the sound propagation conditions. The sound level at most distances is reduced with more than 30 dB from 06 UTC to 12 UTC. Predictions based on the observed atmospheric profiles at 12 UTC show relatively higher sound levels to the north of the source. This is due to the fact that that there was a weak wind (1-2 m/s in the lower 100 m above ground) from the south which was not predicted by MM5 (Figure 5). This is a good demonstration of how sensitive acoustical conditions are to small changes is the local weather pattern.

5. Model changes

The energy balance at the surface is given by:

(2)

where a is the surface albedo, S¯ is short wave radiation, L is downward and upward long wave radiation, H is sensible heat-flux, L× E_tot is latent heat-flux, G₀ is heat-flux down in the soil and D Q_s is storage of energy in the surface. The radiation scheme in MM5 has been changed in order to take in to account the effect of sloping terrain in the calculation of irradiance at the surface (S). The changes are based on Skartveit & Olseth (1986) and Skartveit & Olseth (1987).

According to Skartveit & Olseth, (1987), knowing the hourly diffuse and beam irradiances, S_d and S_b, on a horizontal surface, the total irradiance on a surface inclined by an angle b towards an azimuth angel g can be written:

(3)

where h is solar elevation and q is the solar beam's angle of incidence. Negative q is replaced by zero.

Examples of net short wave radiation are shown in Figure 6 It is clearly seen how dependent the radiation is on the orientation and slope of the terrain (A) compared to flat grid-boxes (B), as assumed in the original radiation scheme. Similar patterns can be seen in the temperature at 2 meter and the sensible heat-flux (not shown). There is an effect on the vertical thermal structure as well although a bit smaller than at the surface. It should be stressed that these results are preliminary.

More tests with other surface conditions, cloud cover and other seasons should be performed before conclusions on the usefulness of the modifications are made. In addition, how the modifications will act with coarser vertical and horizontal resolution should be investigated.