A Look at the Causes of Four Underforecast Heavy Snow Events in the Lower Ohio Valley in the Winter of 2010-11
·        By Mike York, Robin Smith, Pat Spoden, Christine Wielgos, Kevin Smith, Dan Spaeth, Ryan Presley, Gregory Meffert, and Jayson Wilson
·        NOAA/National Weather Service Forecast Office, Paducah, KY
 
 
    During the very active winter of 2010-11, a series of heavy snowfall events caused significant disruption of transportation and commerce in western Kentucky and southeast Missouri. Four of these heavy snowfall events were investigated closely in the following months. Each of these four particular events were selected for further review because they contained unusually heavy snowfall rates (>2.5 cm/hour) that were underforecast by computer guidance and meteorologists alike.
 
     The general goal of these case studies was to assess model performance and to raise forecaster awareness of model deficiencies. In particular, the researchers focused on snowfall intensities and which tools worked best for anticipating heavy snowfall rates (at least 2.5 cm/hour). The influence of snow to liquid ratios and their forecast accuracy was examined in depth.
 
     The primary finding was that mesoscale banding of precipitation echoes played a major role in all of the cases studied. Additionally, the potential for banded snow could have been anticipated by examining model depictions of various derived fields. The only two models reviewed in this study were the GFS and NAM (40 km or lower resolution).
 
    All of the underforecast events were associated with mesoscale banding that resulted in unexpectedly heavy snowfall rates. In each of the four cases, the presence of banding was suggested by model depictions of Petterson Frontogenesis in the 850 to 700 hPa layer. Above this layer at 500 hPa were slightly negative values of equivalent potential vorticity (EPV), which suggested that upward vertical motion would be sustained through a relatively deep layer. Finally, the operational computer models’ depiction of q-vector convergence in the 500 to 300 hPa layer indicated an organized area of upper forcing during all four events.
 
     The mesoscale bands were not explicitly resolvable by the low-resolution operational forecast models. However, the forecaster could have relied on model depictions of frontogenesis, EPV, and upper-level q-vector convergence to diagnose the potential for mesoscale banding and locally heavy snowfall.
 
     The event-total quantitative precipitation forecast (QPF) from the operational GFS was underdone in three of the four events. This was true even within 6 to 12 hours of the onset of precipitation. This would not be unexpected given the inability of the model to explicitly forecast mesoscale bands. The model QPF amounts in the fourth event were rather close, but the axis of the heaviest amounts was oriented more north-south than the model depicted. In the future, when forecasters recognize the potential for banded snow, the model QPF may need to be adjusted upward.
 
     An entirely separate aspect of the study focused on snow-to-liquid ratios. The intention was to determine the forecast accuracy of these ratios. Forecasters often subjectively assume a snow-to-liquid ratio of 10 to 1 with wet snows falling near freezing. Drier powdery snows are assumed to be closer to 20 to 1 or higher. Based on the limited dataset produced by this study, it appears that forecasters may be underestimating snow-to-liquid ratios in wet snow situations. 

    Three of the four events occurred with surface temperatures within two degrees Celsius of freezing. As expected, ratios were relatively low in these situations, but not as low as the assumed 10 to 1 ratio. The average ratio in these situations was closer to 15 to 1 than 10 to 1. A look at model time-height cross-sections of vertical velocity (omega) and temperature revealed rather strong upward velocities in the prime dendritic growth zone region (minus 8 to minus 13 Celsius). In the future, forecasters might benefit from a closer examination of omega in the dendritic growth zone. Given surface temperatures near freezing, snow-to-liquid ratios may be assumed to average close to 15 to 1 when conditions are favorable for ice crystal growth. During the event that occurred in an arctic air mass, forecasts were close to the assumed 20 to 1 ratio.

Event #1 - December 24, 2011

snowfall on Dec. 24

snow to liquid ratios for Dec 24

qg forcing and mslp on Dec. 24

500-300 hPa layer Q-vector conv., MSLP (NAM model, Dec. 24 18z T+6 hours) 

qg forcing and mslp on Dec. 24

500-300 hPa layer Q-vector conv., MSLP (NAM model, Dec. 24 18z T+12 hours)

frontogenetic forcing on Dec. 24

850-700 hPa layer Pettersen Fn, 500 hPa EPV (dashed blue=negative EPV) (NAM model, Dec. 24 18z T+6 hrs)

frontogenetic forcing on Dec .24

850-700 hPa layer Pettersen Fn, 500 hPa EPV (dashed blue=negative EPV) (NAM model, Dec. 24 18z T+12 hrs)

time height of omega on Dec. 24

Time-height cross-section of omega, temperature (NAM model Dec. 24 18z - Over Paducah, KY)

temp lapse rates and rh time-height cross-section on Dec. 24

Time-height cross-section of RH, temperature lapse rates (NAM model Dec. 24 18z - Over Paducah, KY)

Dec .24 radar

Image from Paducah radar at 6:30 P.M. CST Dec. 24. Banded snowfall is circled.

A weak surface low pressure system moved eastward across the Lower Mississippi Valley on Christmas Eve as a rather strong upper level disturbance moved southeast across the central Plains. On radar presentations, the cellular appearance of the heaviest snow indicated there was convective enhancement. Much of the snow fell in a three-hour window, when visibility dropped to one-half mile and accumulations neared an inch per hour. Numerous accidents were reported on this holiday evening.

Event #2 - January 25-26, 2011

Jan. 25-26 snowfall

Snow to liquid ratios on Jan. 25-26

Q-vector convergence and mslp on Jan. 25

500-300 hPa layer Q-vector conv., MSLP (NAM model, Jan. 25 18z T+6 hours) 

Q-vector convergence and mslp on Jan. 26 at 06z

500-300 hPa layer Q-vector conv., MSLP (NAM model, Jan. 25 18z T+12 hours)

Pettersen frontogenesis and 500 mb epv on Jan. 25

850-700 hPa layer Pettersen Fn, 500 hPa EPV (NAM model, Jan. 25 18z T+6 hrs)

Frontogenesis and 500 mb epv on Jan. 25-26

850-700 hPa layer Pettersen Fn, 500 hPa EPV (NAM model, Jan. 25 18z T+12 hrs)

Omega and temp profiles on Jan. 25-26

Time-height cross-section of omega, temperature (NAM model Jan. 25 18z - Over Madisonville, KY)

Profile of lapse rates and rh for Jan. 25-26

Time-height cross-section of RH, temperature lapse rates (NAM model Jan. 25 18z - Over Madisonville, KY)

Radar image from evening of Jan. 25

Image from Paducah radar about 9:30 P.M. CST Jan. 25. Banded snowfall is circled.

 
An upper level storm system moved from western Arkansas across Tennessee. The heaviest snow occurred in bands that pivoted around the upper level system. Snowfall rates in the bands were around an inch per hour. Roads became very slick and hazardous. There were numerous accidents around the region. On the morning following the snowstorm, a fatal wreck was caused by lingering patches of snow and ice on Interstate 24 near exit 40 in Lyon County. The collision caused the westbound lanes of the interstate to be shut down.

Event #3 - February 7, 2011 

Snowfall amounts on Feb. 7

Snow to liquid ratios on Feb. 7

Q-vector convergence and mslp on Feb. 7 at 6 A.M. 

500-300 hPa layer Q-vector conv., MSLP (GFS model, Feb. 7 12z  init.) 

Q-vector convergence and mslp on Feb. 7 at noon 

500-300 hPa layer Q-vector conv., MSLP (GFS model, Feb. 7 12z T+6 hours)

Frontogenesis and epv on morning of Feb. 7 

850-700 hPa layer Pettersen Fn, 500 hPa EPV (GFS model, Feb. 7 12z init)

Frontogenesis and epv at noon on Feb. 7 

850-700 hPa layer Pettersen Fn, 500 hPa EPV (GFS model, Feb. 7 12z T+6 hrs)

Time-height cross section of omega and temp 

Time-height cross-section of omega, temperature (GFS model Feb. 7 12z - Over Marshall Co., KY)

 Time-height cross-section of rh and temp lapse rates

Time-height cross-section of RH, temperature lapse rates (GFS model Feb. 7 12z - Over Marshall Co., KY)

Radar image from mid-morning on Feb. 7

Image from Paducah radar about 9:30 A.M. CST Feb. 7. Banded snowfall is circled.

A surface low pressure system developed in the Lower Mississippi Valley, then strengthened as it moved northeast across TN and KY. In the upper levels of the atmosphere, a rather strong disturbance tracked northeast from eastern AR toward the Lower Ohio Valley. Snowfall rates were measured up to 2 inches per hour. Roads became snow-covered and slippery. Numerous vehicles slid off roads or crashed during the heaviest snow. Many schools decided to open at the normal time, but the snow became heavy before busses even reached the schools. The students were sent back home almost as soon as they arrived at school. The snow disappeared rather quickly from paved surfaces a few hours after the snow ended, due to relatively warm pavement temperatures.

 

Event #4 - February 9, 2011

Feb. 9 snowfall map

snow-to-liquid ratios for Feb. 9

Upper-level q-vector convergence, MSLP

500-300 hPa layer Q-vector conv., MSLP (GFS model, Feb. 9 12z init.) 

Q-vector convergence and MSLP for Feb. 7 noon

500-300 hPa layer Q-vector conv., MSLP (GFS model, Feb. 9 12z T+6 hours)

Frontogenesis and EPV on morning of Feb. 9 

850-700 hPa layer Pettersen Fn, 500 hPa EPV (GFS model, Feb. 9 12z init.)

Frontogenesis and epv on Feb. 9 at noon 

850-700 hPa layer Pettersen Fn, 500 hPa EPV (GFS model, Feb. 9 12z T+6 hrs)

Time-height cross-section of omega and temp 

Time-height cross-section of omega, temperature (GFS model Feb. 9 12z - Over Murray, KY)

Time-height cross-section of rh and temp lapse rates 

Time-height cross-section of RH, temperature lapse rates (GFS model Feb. 9 12z - Over Murray, KY)

Paducah radar image on Feb. 9 late morning

Image from Paducah radar about 10:50 A.M. CST Feb. 9. Banded snowfall is along Tennessee/Kentucky border.

A large area of snow occurred along and to the east of a well-defined inverted surface trough. The heaviest snow occurred just ahead of a disturbance in the mid levels of the atmosphere. Snowfall rates in a narrow band near the TN border were 1-2 inches per hour. Due to very cold temperatures, roads quickly became snow-packed and very slippery. Numerous accidents were reported. Visibility was 1/4 mile or less in the heavy snow areas. Following the snow, a combination of residual moisture and warm river water, dense freezing fog formed and caused a fatal wreck on the Interstate 24 bridge westbound over the Cumberland River. As a tractor-trailer rig approached the crash site, the accident victim jumped off the bridge to avoid being struck and died.

 


USA.gov is the U.S. government's official web portal to all federal, state and local government web resources and services.