Columbia Canal Breached by Congaree River Flooding in South Carolina

This issue’s cover photo is a view north up the Columbia Canal (Columbia, South Carolina) from near the Gervais Street Bridge on October 14^th 2015, after it was dammed off with a rock barrier (upper right). The canal was dry due to a breach in the levee separating the canal from the Congaree River, sometime on October 4^th after being over-topped by high water.

Much of the local and national news coverage of the October 2015 Columbia floods focused on high water along Gills Creek, a small urban watershed on the east side of Columbia that drains into the Congaree River floodplain about five miles below the canal. Gills Creek has a large number of reservoirs and ponds, unlike many of the smaller urban watersheds that drain downtown Columbia. These water bodies usually store flood water and reduce flood peaks. The unprecedented intensity and duration of this precipitation event, however, led to dam failures and the subsequent dam-burst floods propagated downstream. At least four dams in the Gills Creek Watershed failed: Cary’s, Upper Rockyford, Lower Rockyford, and Semmes Lakes. Flooding was highly destructive but could have been much worse if the larger dams holding Forest Lake and Lake Katherine had failed. Forest Lake Dam was overtopped by swift flows across the entire length of its crest, but it held.

How extreme were these precipitation and flood events?

The Congaree River near Columbia streamflow gauge—1.5 km (1 mi) downstream of the Canal breach—has a very long continuous flood record. While the recent event was not a record flood on the Congaree, it was the largest in 79 years.

Gills Creek streamflow probably has a very long return period, because it was not only generated by an extreme rainfall, but it was also augmented by dam-burst flows from multiple reservoirs. The peak stage was not recorded directly due to loss of the gauge at Gills Creek.

Unlike streamflow, rainfall intensities were unprecedented at many locations. All-time records for one-day, two-day, and three-day rainfall amounts were set in the Columbia area (NWS, 2015a). On the east side of Columbia, the Gills Creek weather station recorded 546 mm (21.49 inches) of rain from Friday through Monday (NWS, 2015a), exceeding the 1000-year three-day rainfall event by 137 mm. In comparison, the three-day precipitation from the University of South Carolina (USC) weather station on campus near downtown Columbia was 299 mm (11.76”) (NWS, 2015b), a greater intensity than a 200-year three-day rainfall event.

The calculation and interpretation of rainfall and flood frequencies is complicated by methodological and philosophical difficulties. The 2015 flood on Gills Creek was greatly influenced by the failure of ponds and reservoirs and it is not clear if this risk was factored into flood risk assessments and base-flood mapping. The assumption of stationarity in hydrology, i.e. unchanging hydroclimatological probabilities through time, underlies most conventional magnitude-frequency computations. However, this assumption is difficult to make due to changes in probabilities of both rainfall and streamflow. Some hydrological scientists have pronounced stationarity ‘dead’ as the default operational assumption for water-resources risk assessment and planning (Milly et al., 2008).

This difficulty in quantitatively estimating flood risks is manifested in the 2015 Columbia, SC flood. Moreover, as climate warming increases temperature and pressure gradients between the equator and the poles, global circulation patterns will become increasingly meridional as opposed to westerly, and this and other changes will increase the variability of weather systems. Put simply, flood destruction as well as economic losses to drought are likely to increase. We in Columbia recently received this message loud and clear.