Figure 1.
Topography and bathymetry of the Lake Erie basin.
The lake is characterized by three major basins from west to east that are distinguished by their depth. There are NOAA measuring stations at the Fermi power plant, Toledo, and Buffalo.
Table 1.
Dimensions of the Lake Erie domain.
Figure 2.
Wind forcing at Toledo during December 1–2, 2006.
Panels: (A) Toledo wind speed, (B) Toledo wind direction. The lines labeled "RUC" represent wind data from NOAA's Rapid Update Cycle interpolated from their 20 km grid onto the 1 km Lake Erie domain (Section 2.2).
Figure 3.
Wind forcing at Buffalo during December 1–2, 2006.
Panels: (A) Buffalo wind speed, (B) Buffalo wind direction.
Figure 4.
Wind forcing at Toledo during January 30–31, 2008.
Panels: (A) Toledo wind speed, (B) Toledo wind direction. The lines labeled "RUC" represent wind data from NOAA's Rapid Update Cycle interpolated from their 20 km grid onto the 1 km Lake Erie domain (Section 2.2).
Figure 5.
Wind forcing at Buffalo during January 30–31, 2008.
Panels: (A) Buffalo wind speed, (D) Buffalo wind direction.
Figure 6.
Ice coverage on Lake Erie, winter 2008.
Panels: (A) 28 January 2008, (B) 31 January 2008. The windstorm cleared much of the ice from Lake Erie. These images were created using archive data from the Canadian Ice Service, Government of Canada.[11]
Figure 7.
Significant wave height on Lake Erie in steady-state (experiment S1).
SWAN generated these waves driven by winds at 20 m/s blowing from the southwest and aligned with the long axis of the lake (20° north of due east).
Figure 8.
Maximum wave heights on Lake Erie for constant wind speeds (steady-state experiment S2).
Wave height increases with the square of the wind speed until about 15 m/s, when wave growth is constrained by the lake's limited depth.
Figure 9.
December 2006: Wind setdown and storm surge, and simulation experiments E1–E4.
Panels: (A) Fermi 2006, (B) Buffalo 2006. E1: ROMS in 2-dimensional mode. E2: 3-D mode with 3 vertical levels. E3: SWAN-generated waves and nearshore radiation stress (NEARSHORE_MELLOR08).[33] E4: SWAN-generated waves and vortex force.[35],[34] Line E3 lies between E1 and E2. Vortex force causes waves to have a larger effect than other the methods.
Figure 10.
December 2006: Wind setdown and storm surge, with experiments E16–E17.
Panels: (A) Fermi 2006, (B) Buffalo 2006. E16: 3 vertical levels. E17: 10 vertical levels. Increasing the number of vertical levels from 3 to 10 causes only a small increase in wind setdown and storm surge for this case.
Figure 11.
December 2006: Wind setdown and storm surge, with experiments E4, E18, E21, and E23.
Panels: (A) Fermi 2006, (B) Buffalo 2006. E4: RDRG2 = 3.0e-3. E18: RDRG2 = 1.0e-3. E21: Vortex force * 0.8. E23: Adjusted the modeled wind speed at Buffalo. Decreasing the bottom drag RDRG2 improves the timing of the model by shortening the lake's reaction time to changes in wind stress (E4 -> E18). Decreasing the vortex force corrects the overshoot of setdown and surge (E18 –> E21). Simple data assimilation at Buffalo provides a further small improvement (E21 –> E23).
Figure 12.
January 2008: Wind setdown and storm surge, with experiments E25, E26 and E31.
Panels: (A) Fermi 2008, (B) Buffalo 2008. E25: Drag coefficient Cd * 1.3. E26: Cd * 1.1. E31: Cd * 1.0. Lake ice increased the effective drag coefficient by a factor of 1.1.
Table 2.
Modeled wind setdown and storm surge (m).
Table 3.
Comparison of drag coefficients.
Figure 13.
Corrections applied to the Lake of Tanis and the Kedua Gap.
Reducing the quadratic bottom drag increases the duration of the dry land bridge, as does the inclusion of waves in the ocean model. T23: original configuration using RDRG2 = 3.0e-3 and no waves. T21: RDRG2 = 1.0e-3 and no waves. T24: RDRG2 = 3.0e−3 and SWAN-generated waves. T25: RDRG2 = 1.0e−3 and SWAN waves.