The snowbands of winter storms

Many of us have experienced the nervous anticipation that precedes the arrival of a potentially major winter storm. We fill our gas tanks, stock up on necessities, and wait to hear of school or work closures. Then it barely snows in our community, though locations nearby get buried. We’ve just experienced the effect of snowbands—linear regions of enhanced snowfall rates embedded within surrounding lighter precipitation.

Many studies have focused on large (at least 200 km in length), single snowbands, yet less has been done to understand smaller-scale bands. Such bands tend to exist in groups and are capable of producing heavy snowfall and wind gusts. The Boxing Day Blizzard of 26–27 December 2010, for example, effectively shut down an ill-prepared New York City with its strong single band and multibands. Analysis of the frequency of multibands within East Coast winter storms and their impact, as compared with single bands, can improve forecasting of snowfall amounts. In a recent study, my colleagues and I identified and classified snowband structures in more than 100 winter storms. By comparing single-band and multiband atmospheric environments, we took an important step toward learning why winter storms produce one rather than the other.

Snowbands typically form northwest of a developing surface low-pressure center, as shown in figure 1a/1b, in a region of increasingly strong temperature gradients and ample moisture. Air parcels rise within such a frontal zone, which leads to enhanced precipitation. The strongest observed snowbands typically form within frontal zones that are characterized by slantwise instability, in which air is able to rise vertically at an angle, as shown in figure 1c/1d.

To analyze snowbands, my colleagues and I identified 108 cool-season (October–April) low-pressure systems that tracked through the Northeast US and dumped at least 0.75 inch liquid equivalent of snowfall in the New York City metropolitan area from 1996 to 2016. We combined radar reflectivity data from six sites into a single, continuous, 2-km-above-ground-level data set and used software to identify features, or objects, of higher reflectivity in that data. An object was determined to be a snowband if its aspect ratio (length of minor axis divided by length of major axis) was no more than 0.5. Bands were additionally classified as large bands, midsize bands, or undefined, according to the criteria shown in the table below.

Each of the 108 winter storms was given one of four classifications based on which types of objects were identified: single band (SINGLE), multiband (MULTI), both single band and multiband (BOTH), and nonbanded (NONE). The SINGLE category is for a storm that exhibits only a large band. A MULTI storm exhibits at least two midsize bands that have similar orientations and motions. The BOTH category includes storms that exhibit a large band accompanied by midsize bands that have similar motions. Storms that exhibit none of the aforementioned criteria were tagged NONE. Examples of cases in each of the four categories are found in figure 2.

Exactly half of the 108 storms fit into the BOTH category. Contrary to previous studies, our inquiry found that multibands typically accompany a large band at some point during its life cycle; they were absent in only two storms that we studied. The location of the clusters for each classification can be seen relative to the cyclone center, or low-pressure minimum, in figure 3. All three banded classifications tend to occur in the northern quadrants relative to a cyclone center, whereas the NONE classification also extends to the southeast. We also analyzed the storms to compare 56 stronger, mature storms with 54 weaker, developing storms. We found that developing storms mostly fit the NONE classification (22) or the BOTH classification (26), whereas mature storms were mostly associated with the BOTH classification (33).

We then set out to explore the atmospheric conditions near the snowbands for various kinds of snowstorms. Vertical profiles of temperature, moisture, and winds were taken at the sounding balloon location closest to the bands during the times of subjectively determined peak banding activity.

In examining the thermal profiles, we found insignificant differences among single-band, multiband, and nonbanded environments. That could be the result of the fact that temperature and moisture characteristics are similar in the northern quadrants of winter storms. However, calculations of the horizontal temperature gradient showed that the MULTI and NONE classifications were associated with weaker values than the SINGLE and BOTH classifications. That was true for both mature and developing cyclones, which supports the idea that larger, single bands result from frontal air motions but smaller bands do not.

We are currently working toward using the high-resolution Weather Research and Forecasting model to simulate several winter storm cases, including the BOTH case of the 2010 Boxing Day Blizzard. The main challenge when attempting to re-create fine-scale snowbands with a weather model is determining a sufficient horizontal grid spacing. If the grid spacing is too coarse, the model might not be able to accurately resolve the snowbands. If the grid spacing is too fine, the model may break a snowband down into individual, chaotic circulations that resemble clusters of thunderstorms.

The analysis of snowbands within Northeast US winter storms is an important research topic because of the high-impact nature of the storms. Meteorologists need to better understand the spatial and temporal scales of snowbands to improve snowfall prediction. Weather forecasters rely on high-resolution weather models that can resolve both small-scale multibands and large-scale single bands. Therefore, it is imperative that multibands are understood as realistic and observable structures that can result in localized enhanced snowfall rates. Emerging technologies such as dual-polarization radars, phased-array radars, and snowflake cameras, along with data from high-resolution numerical weather models, can provide greater insight into the structures and environments of different banding classifications.

Sara Ganetis received her PhD in marine and atmospheric sciences and is currently a meteorologist-developer with I. M. Systems Group Inc working for the Weather Prediction Center, a national arm of the National Weather Service. This research by Ganetis was part of her dissertation work under a project managed by Brian Colle of Stony Brook University and Sandra Yuter of North Carolina State University, with considerable collaboration on radar data from Nicole Hoban of North Carolina State University.