WRF simulations of thunderstorms triggered by the Alberta oil sands development

The Athabasca oil sands development in northeastern Alberta has created a dramatic disturbance to the natural boreal forest ecosystem. More than 500 km² of boreal forest has been converted into an amalgamation of roads, barren ground, tailings ponds, and bitumen upgrading facilities. Research suggests that cities and industrial complexes can affect thunderstorm initiation timing and frequency, as well as lightning and precipitation amounts. Our previous research shows that the oil sands development has not significantly affected lightning or precipitation climatology, but has created a heat island and a dry island, both of which are increasing in intensity as further oil sands development takes place. For this research, a numerical model was used to investigate whether the oil sands development could affect thunderstorm initiation and intensity on one case study day rather than climatologically.

The WRF-ARW (Weather Research and Forecasting Advanced Research WRF) numerical weather model was used to carry out sensitivity studies on a past thunderstorm case near the oil sands development. We are interested in whether land cover changes and increases in surface heat flux caused by the oil sands development could affect thunderstorm initiation and intensity. A factor analysis with two factors (Stein and Alpert 1993) was used to highlight the differences between the four model simulations. The land use was changed from the original boreal forest to barren land, and 100 Wm^-2 of industrial heat was added to the atmosphere over 10% of the oil sands land area. Most of the industrial waste heat is added by industrial activity concentrated in small areas. The case with barren ground and industrial heat added to 10% of the land area represents reality the closest, and we expect it to simulate the weather the closest also.

On the case study day of July 29, 2010, there was plenty of instability and a weak capping inversion. A slow moving thunderstorm was initiated at 1830Z over the oil sands development and traveled southeast giving 21.0 mm of rain to the Fort McMurray Airport between 1910Z and 1940Z.

1) Removing Boreal Forest

The results show that changing the land use from the natural boreal forest to barren ground slightly reduced the thunderstorm intensity, but triggered thunderstorms earlier. The areal average of the total accumulated precipitation decreased from 2.3 mm to 2.1 mm, and the maximum radar reflectivity decreased from 49.4 dBZ to 48.0 dBZ. However, the storms near the oil sands development were triggered a full hour earlier than in the case without any land disturbance. Removing the vegetation raises the Bowen ratio because there is more sensible heating and a lower latent heat flux. Less intense thunderstorms are likely caused by decreased humidity caused by lower evapotranspiration; however, the earlier initiation is likely caused by the increased sensible heating.

2) Adding Industrial Heat

Adding concentrated industrial heat to the natural boreal forest slightly increased the storm intensity and triggered thunderstorms earlier. The areal average of the total accumulated precipitation increased from 2.3 mm to 3.2 mm, and the maximum radar reflectivity increased from 49.4 dBZ to 50.2 dBZ. Storms near the oil sands development were triggered 45 minutes earlier than the case without any industrial waste heat. Adding extra heat should trigger storms earlier, and because the same amount of moisture is available, it is not surprising that the storms are slightly stronger.

3) Removing Boreal Forest and Adding Industrial Heat

We believe that the most realistic representation of the heat and moisture fluxes from the oil sands development was when we removed the boreal forest and added concentrated industrial heat at the same time. The areal average of the total accumulated precipitation increased from 2.3 mm to 2.8 mm, and the maximum radar reflectivity increased from 49.4 dBZ to 49.9 dBZ. However, storms were triggered 1.5 hours earlier than the case without any land disturbance. In this case, storms were triggered much earlier, and were slightly stronger than in the case without any land disturbance. However, they were slightly weaker than the case with just the forest removed. Turning on both factors resulted in a smaller impact than the sum of the two individual factors.

4) Discussion

Small variations in thunderstorm intensity were detected between the 4 model runs. Generally, removing the trees to leave barren land resulted in earlier, less intense storms. Adding a concentrated industrial heat flux with or without clearing the trees resulted in earlier, more intense storms. However, even the largest intensifications of thunderstorms were relatively minor. In contrast, changing the land use and industrial heating had a much more significant effect on the timing and location of thunderstorm initiation. With both factors activated, thunderstorms were initiated 1.5 hours earlier over the oil sands. The main mechanism seems to be a plume of heat from the oil sands development helped to trigger thunderstorms by breaking through the capping inversion. Thunderstorm initiation on this day matched reality closest when the oil sands development was present in the model. It deviated significantly from reality in the boreal forest case. Although there were minor thunderstorm intensity variations caused by the oil sands development, it seems that the most significant impact is on the timing of thunderstorm initiation. This case most closely matched real data if the oil sands development was present in the model. The oil sands development may not yet be big enough to drastically change the thunderstorm intensity, however further research should investigate how a future larger oil sands project could impact thunderstorm intensity. The areal extent of the oil sands development is forecast to double in the next 10 years.