The Washington PostDemocracy Dies in Darkness

Kansas hit with near-record blast of snow that weather models didn’t see coming

More than 2 feet of snow fell in a narrow swath of Colorado and Kansas on Tuesday

The Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA's Terra satellite captured this false-color image (bands 7-2-1) of snow over the Colorado-Kansas border Tuesday. Snow appears in turquoise, while clouds appear white. (NASA EOSDIS LANCE, GIBS/Worldview)
Placeholder while article actions load

An unusual band of localized heavy snow unleashed over 2 feet of accumulation Tuesday on a narrow swath of Colorado and Kansas, nearly breaking the latter’s all-time snowfall record.

The swath’s existence and placement were modeled and officially predicted, though the amount of snow that accumulated far exceeded even the most aggressive estimates. It was an event that exemplified the difficulty of forecasting small-scale snow banding, an ever-present challenge in winter that’s accentuated by the tendency of such storms to have serious consequences.

A lobe of cold, dense upper-atmospheric air, known as a shortwave, swung south into the Great Basin on Tuesday. A wide vacuum of low pressure developed around 10,000 feet aloft, centered over Colorado. Air racing toward this region of low pressure, high above the ground, blew quickly from the southeast to northwest over the Kansas-Colorado border.

Major winter storm, possible blizzard to slam Northeast on Saturday

But it encountered an obstacle: air furiously rounding a much larger mid-atmospheric zone of low pressure over the Hudson Bay, flowing northwest to southeast. Where the former smacked into the latter, air had nowhere to go but up.

As air rose along the mid-level front, it triggered moisture sourced from the Pacific Ocean to condense into precipitation. A narrow but intense band of snow developed, with snowfall rates exceeding 2 inches an hour under its core. Both the Colorado low-pressure zone and the Hudson Bay low remained largely immobile during the day Tuesday, and the mid-level wind-shift placement varied little as a result. This kept the snow band in the same place for about nine hours, delivering a thin swath of extreme snow totals.

Several locations saw accumulations that approached or even exceeded 2 feet. Mount Sunflower, Kan., reported 27 inches of snow, which came within three inches of the state record. Just miles away, substantially less snow fell, testament to the narrow width and limited movement of the long-duration band.

High-resolution forecast models had depicted the likelihood of such a band of snow developing as much as a day before precipitation began to fall. Hours before the event, models even placed the band within miles of where it would end up developing. But even these simulations failed to resolve the magnitude of accumulations, with maximum snow totals placed below 15 inches.

The National Weather Service office in Goodland, Kan., also predicted the band’s existence and placement well, but under-forecast the maximum totals. The office issued winter storm warnings for the areas affected, and cited in its 3 a.m. forecast discussion the potential for “an intense snow band developing around sunrise and persisting through the afternoon hours with 1-2 inch per hour snowfall rates; snowfall amounts in this band may approach 8+ inches in spots.”

Computer models are guidance, not gospel, a reminder to forecasters and their users

The under-forecasting of snowfall magnitude is a testament to challenges in resolving this sort of small-scale banding, both by models and human forecasters. It is a difficulty that extends to all snowstorms intense enough to feature intense bands, representing a high-impact but low-confidence hazard. As a major storm threatens the Northeast this weekend, similar challenges will probably face forecasters there.