A preliminary climatology of these cold surges was constructed by calculating the one-hour temperature change for the Great Falls ASOS from 1998 to 2014. Surface analyses and station models were then examined to separate the true cold surges from other causes of abrupt cooling. To obtain 20+ cases, an hourly temperature change of 16 degrees Fahrenheit (8.9 degrees Celsius) was used as an event threshold. Generally, between zero and three cold surges occur each year with a maximum in January and a secondary maximum in late summer-early fall. Very few events occurred between the hours of 19-23 UTC (12-16 LST), which is significant because it is only in this timeframe that a lag in the model forecast will affect the typical time of the maximum temperature.
As an example of this model lag, the cold surge of 27-28 January 2008 was simulated using the WRF model with one nested 4-km domain. While the model captures the structure of the cold surge very well, the timing is slow by approximately 2.5 hours. Sensitivity tests with other PBL schemes result in variations in this timing (and the strength of the cold surge as well), but these are not sufficient to bring the results in agreement with reality. Additionally, altering the surface roughness in certain grid cells in southern Alberta produced a more realistic wind field, but did not accelerate the cold surge significantly.