2.0 MC2 MODEL OUTPUT COMPARED TO OBSERVATIONS
To evaluate the performance of the MC2 model under conditions dominated by thermally-driven mesoscale circulations, a day with well-developed lake breezes relatively unperturbed by the gradient flow was required. Meteorological conditions on August 8, 1993 met these criteria and simulations of this day's conditions are used as the baseline run. A synoptic scale high pressure system was present over the region on August 8 preventing any significant convection from developing. Also, data from the Hastings, Ontario site have already been analyzed for this day. At Hastings, the Lake Ontario lake breeze front passed near 1820 EDT. Ozone concentrations at the Hastings site increased sharply from 48 ppb to 59 ppb as the front passed.
To simulate the evolution of the Lake Ontario lake breeze on this day, we are using a 100 x 100 horizontal mesh with 5 km grid spacing. Vertically, we use 28 levels with the lowest level at 10 m and the highest level at 24389 m. To provide initial and boundary conditions for this simulation, model runs are needed at 100 km grid-spacing and 25 km grid-spacing with the output used as nesting data. Initially, cases were modelled using version 3.2 of the model. However, we began using version 4.0 of the model in August of 1997. Improvements to the model physics in version 4.0 made this change desirable. Version 4.1 of the model became available shortly after the release of version 4.0, but due to problems with the new code, we were unable to use version 4.1 with the enhanced vertical resolution we required in the boundary layer. Thus, the simulations carried out for this experiment used version 4.0 of the MC2 model.
2.1 Coarse Mesh Model Runs
In order to provide initial and boundary conditions for the fine mesh model run with 5 km grid spacing, two coarse mesh runs are required. The first coarse mesh model run is performed with 100 km grid spacing on an 87 by 59 node mesh using a 20 minute time step. Twenty-eight vertical levels are used for both the coarse mesh and fine mesh runs with more than one half of the levels located below 3000 m (above ground level), just above the typical maximum extent of the lake breeze circulation. The second coarse mesh model run is performed with 25 km grid spacing on a 84 by 70 node mesh using a 5 minute time step. The following sections compare the output from these coarse grid runs with objective analyses and observations.
2.1.1 100 km Grid Run
This run, with a domain centred over Canada and the United States, had initial and boundary conditions provided by CMC Regional Finite Element model (the operational model used in 1993) objective analyses valid at 00Z and 12Z on August 8, 1993 and 00Z on August 9, 1993. The integration was carried out for twenty-four hours with selected fields being output every hour. Figure 2.1 shows the surface pressure and 10 m winds valid at the end of the integration, 00Z August 9. Figure 2.2 shows the same fields taken from the objective analysis at that time. Over the eastern half of North America, including the Great Lakes region, the pressure and wind patterns are very similar with a dome of high pressure dominating the eastern Great Lakes. Both plots show a central value of 1026 mb within this region of high pressure, though the model forecasts the highest pressure to be just offshore from New England while the objective analysis places the highest pressure just south of Buffalo, New York. On the western side of the continent the pressure and wind patterns differ significantly. The objective analysis places a 1003 mb low over central Saskatchewan and a 1018 mb high over Colorado while the model predicts a 999 mb low over northern Saskatchewan and a 1002 mb low over Nevada. The associated wind patterns at the surface thus differ substantially. It seems that the behaviour of the pressure solver may be problematic over the Rocky Mountain regions. However, since our region of interest has relatively modest topography, we have not concerned ourselves with this aspect of the model.
2.1.2 25 km Grid Run
Using initial and boundary conditions from the 100 km grid run, the 25 km grid run was centred over the Great Lakes basin. The integration was carried out for 18 hours from 06Z on August 8 to 00Z on August 9 with selected fields output every hour. Figure 2.3 shows the AES surface analysis valid at 00Z on August 9. A dome of high pressure was anchored over the eastern Great Lakes area with a central pressure of 1027 mb southwest of Buffalo, New York and another central pressure of 1026 mb along the New York - Vermont border. Weak frontal systems to the north and south of the Great Lakes area were generating showers and thunderstorms. Winds were generally anticyclonic around the dome of high pressure. A stronger pressure gradient existed west of the Great Lakes. The model surface pressure and surface winds valid at 00Z on August 9 are shown in Figure 2.4. The model captures the surface pressure and wind pattern quite well near the Great Lakes region and also to the west. However, the highest central pressure produced by the model is only 1025 mb. Also, towards the American eastern seaboard, the 1024 mb contour dips farther south than it does in the analysis.
Observed 1.5 m air temperatures in the Great Lakes basin ranged from the low teens over Lake Superior to the upper twenties south of Lake Michigan (Figure 2.3). Temperatures over southern Ontario were between 19°C and 22°C. Figure 2.5 shows that similar temperatures are produced by the model with values in southern Ontario reaching 23°C. However, the model does produce temperatures near 20°C in parts of the Appalachians which are not observed. Again, the model seems to have the most trouble in areas of mountainous terrain.
The observed winds in the vicinity of the Great Lakes at 00Z on August 9 show widespread onshore flow due to lake breeze activity (Figure 2.3). The model does a good job of simulating these circulations considering the relatively coarse resolution of this run (Figure 2.6). It can be seen that light gradient winds along Lake Erie and Lake Ontario encouraged inland penetration of lake breezes over the extreme western and eastern shores respectively. Lakes Huron and Michigan also produce strong lake breezes.
Figure 2.7 is a remapped GOES-7 visible satellite image valid at 2002Z
on August 8, 1993. The image shows lake breezes in progress over each of
the eastern Great Lakes. The inland extent of the circulation can usually
be identified by a narrow band of cumulus clouds produced at the lake breeze
front. Also, strong subsidence behind the front and over the lake usually
results in clear skies on the front's lakeward side. Figure 2.8 shows that
the model produces a very similar lake breeze pattern at this time.
2.2 Fine Mesh 5 km Model Run
Using initial and boundary conditions from the 25 km grid run, the 5 km grid run was centred over Lake Ontario. Lake Simcoe and small portions of Georgian Bay and Lake Erie are also located within the model domain. Figure 2.9 shows the topography of this region. Note the steep slopes of the Niagara Escarpment to the west of Lake Ontario and the Oak Ridges Moraine that runs along the north shore. The elevation gradually increases to the north while the Appalachian mountains begin to the south of Lake Ontario. The integration was carried out for 12 hours from 12Z on August 8 to 00Z on August 9 using a 1 minute time step with selected fields output at various time intervals. Figure 2.10 shows the modelled wind field valid at 2000Z on August 8. The Lake Ontario lake breeze circulation is the most prominent feature with a meso-high near the mouth of the Niagara River and winds radiating out from this point. Moderate lake breeze winds are evident penetrating up to
15 km inland along the north shore of Lake Ontario. Lake Erie, Georgian Bay and Lake Simcoe also produce lake breeze circulations. Figure 2.11 shows a vertical cross-section through the lake breeze from over Lake Ontario to north of Hastings. Lake breeze onshore flow can be seen on the left side below 200 m with a strong convergence zone at the leading edge of the front reaching up past the 1000 m level. Strong upward and downward motion at the surface is an artifact of the terrain-following coordinate in a region of varying topography.
Observed meteorological parameters such as temperature, dew point and pressure are satisfactorily simulated by the model though there seems to be a problem with modelled dew point at the first model level. Values of dew point temperature exceed temperature values by a significant margin at the first model level in some areas over water, resulting in relative humidity values of 110-115%. Our investigation of the problem resulted in the identification of an error in the model code which is now being corrected by the MC2 group at RPN. However, dew point temperatures over water above this level and those over land away from lakes quickly return to reasonable values. Also, the model shallow convection from the boundary-layer turbulence scheme seems to produce an excess of cumulus clouds over the entire domain in this simulation. Figure 2.12 shows the total cloud cover over the domain in tenths valid at 2000Z. The minimum total cloud cover is found over west-central Lake Ontario with a value near 0.2. The cloud cover over the lake increases to 0.4 at the east end of the lake. Total cloud cover also increases to between 0.4 and 0.5 inland from the lake shoreline. However, visible satellite imagery valid at 2002Z, shown in Figure 2.13, depicts clears skies over Lake Ontario and also at points inland behind the lake breeze front. Varying amounts of fair weather cumulus are observed farther inland. The excess cloud cover produced by the model reduces insolation at the surface and affects the strength of the lake breeze circulation by reducing the land-lake air temperature difference and the depth of the convective boundary layer over land.
Figure 2.14a shows the inland penetration of the Lake Ontario lake breeze through the day of August 8 as inferred using GOES-7 visible satellite imagery. The eastern portion of the circulation could not be detected until 1700Z, but after this time the front could be seen moving steadily inland around the entire perimeter of the lake. According to the satellite imagery, the lake breeze reaches Hastings sometime between 2200Z and 2300Z. The inland penetration of the modelled Lake Ontario lake breeze on August 8 is shown in Figure 2.14b. The lake breeze front was identified using divergence and vertical motion at the 500 m model level as well as 10 m model winds. A well-developed lake breeze front was present along the north shore of Lake Ontario but only managed to penetrate inland about 20 km, or half of the way to Hastings, by 2300Z. The front was also well-developed over the Niagara Peninsula but was much more difficult to identify along the south shore of the lake. No front was discernable along the eastern shore.
Data from the Hastings meteorological station show that the lake breeze front moved past the station between 2220Z and 2230Z. Table 2.1 compares the meteorological conditions before and after the passage of the front. It can be seen that wind speed increases by only
1 knot from 6 knots to 7 knots and that the wind direction shows very little change after the passage of the front remaining out of the west-southwest. However, the temperature decreases by 1.8°C and the dew point temperature increases by 5.2°C. As mentioned previously, large changes in pollutant concentrations, including ozone, were also recorded as the front passed. The observed changes are compared to changes as the model front passed by a grid point to the south of Hastings at the same time (a grid point to the south is used since the lake breeze front never arrived at the Hastings grid point). Across the front, the model predicts a slight increase in wind speed of 1 knot from 4 knots to 5 knots. The wind direction shifts 38° from 259° to 221°. There is a small decrease in the temperature from 22.5°C to 21.8°C and an even smaller increase in dew point temperature from 14.4°C to 14.6°C.
2.3 Summary
Overall, the MC2 simulation of lake breeze conditions on August 8 was less than successful. While the model did manage to produce a lake breeze on Lake Ontario, the strength, penetration distance and frontal characteristics of the lake breeze did not match observations. Basic meteorological variables such as temperature and pressure were modelled satisfactorily, but excessive shallow convection may have helped prevent the lake breeze from evolving as observed. There is a possibility that improved surface parameters such as those being tested in the Sensitivity Testing section may result in a more accurate representation of the Lake Ontario lake breeze on this day. Other possibilities include model misrepresentation of the synoptic scale meteorology at high resolutions or aspects of the various physical parameterizations used in the model.