Looking Back on August 31, 2014 from the Air
Residents of west-central Iowa may remember the powerful squall line that tore across the region on the evening of August 31, 2014, leaving a wide swath of wind damage from Crawford to Dallas counties. More intermittent wind damage was reported from Greene to Howard counties. In the days following the event, the NWS reviewed the wind damage reports/photos and could not conclusively distinguish any damage that could have originated from a tornado. Tornadoes produced by squall lines are oftentimes weak, transient, rain-wrapped, and sometimes embedded within broader straight-line winds, making it very difficult to differentiate tornado versus non-tornado damage in post-storm ground surveys. Mature corn crops can play a vital role in showing the unmistakable convergent path left by a tornado, and two tornadoes from August 31 were located by the author in cropland on a scientific survey on September 10, 2014 (see photo below).
Convergent damage path through corn for one of the ground-surveyed tornadoes on September 10, 2014 just north of Dayton in Webster County.
However, in the weeks following the event, entire 350-km path of the squall line was imaged at ≈1-m resolution using aerial photography through the USDA National Agriculture Imagery Program. As stated earlier, the predominantly flat, mature agricultural land cover of central Iowa provided an excellent medium on which to document all scales of wind phenomena. The imagery (discovered and analyzed during the last 6-8 months) revealed an astounding 111 discrete damage tracks that could have originated from a tornado. These tracks ranged in length from a mere 130 m to nearly 18 km. Given the uniqueness of this dataset and the high likelihood that some of these tracks were from surface vortices that did not meet the formal definition of a tornado (a circulation reaching to cloud-base), a probabilistic testing scheme was developed. This test weighted various track characteristics (length, strength, circulation nature, and damage) and radar data to determine which tracks had the highest chance of being from a tornado. Using this test, 35 of the 111 tracks (31%) were classified as tornadoes. Four of the tracks were rated EF-1 using the aerial data and the rest as EF-0.
Imagery collage of nine tracks from August 31, 2014, highlighting the wide variety of damage patterns observed. Track direction and identification numbers are provided with each event. All images rendered to the same scale.
The sheer number of damage paths revealed by this dataset is truly unprecedented. No other study has ever uncovered so many tracks produced by a single squall line. The aerial data also showed two tornadoes merging into one single entity just northeast of Stratford (see image below), one of only four mergers ever documented and the first with tornadoes from a squall line. This event unearthed far more questions than answers, not only concerning how these tornadoes formed, but also with how the NWS should handle these types of situations. Using hypothetical tornado warnings that would have encompassed the area most at risk to tornadoes on August 31, it was calculated that only 0.24% of the warning area would be impacted by a tornado. Should the NWS be issuing tornado warnings for these low-impact, short-lived tornadoes that would rarely cause damage greater than EF-0/EF-1? Is it worth false alarming over 99.8% of the warning area to try to capture these fleeting tornadoes that would likely go undocumented? In addition, how should the NWS document such tornadoes for historical records? Since aerial data will not be available for all events, a bias would be introduced into Storm Data. The August 31, 2014 squall line exemplifies this bias to the extreme. To put this event in context, the 35 tornadoes from August 31, 2014 would rank as the single greatest tornado outbreak in Iowa history, an outbreak that no one has ever heard of! These questions and more will need to be addressed by the NWS over the coming years not only as the organization moves towards impact-based warnings, but also as the availability and accessibility of aerial/satellite imagery datasets increase. This case highlights the incredible utility of aerial and satellite datasets for storm surveying, a benefit that the NWS will hopefully capitalize on in the years to come.
(a) Polygon paths for tornadoes 62-T19 (red) and 64-T21 (blue) with the start and end times (in UTC) for each tornado annotated. Local streets are provided as a map background. The black dotted outline denotes the region encompassed by the aerial imagery shown in (b) of both track crossing points and the merger, with the estimated times of the first crossing point and merger noted. Tracks are outlined and in the same colors as (a). (c) Close-up imagery of the merger point with different stages of the merging process highlighted.
Blog post by Kevin Skow, Meteorologist Intern, NWS Des Moines