Icy Road Safety.com - Prepare for Weather's Most Underrated Hazard
                Friday, October 29, 2021

Developing a vehicle loss-of-control risk parameter for forecast models and mesoanalysis

RSS/XML feed
Dan's YouTube Video Channel

The purpose of this parameter is to allow quick identification of regions at risk of developing hazardous road conditions due to winter precipitation. The parameter will focus on conditions that pose the highest risk of vehicle loss-of-control incidents.

Factors that increase the risk of hazardous roads

During any type of precipitation, the following factors have an additive effect on the overall risk of hazardous roads:
  • Surface temperatures at or below 32°F OR surface wet bulb temperatures below 32°F. All precipitation types begin to freeze on road surfaces at surface air temperatures or below the freezing mark. Wet bulb temperatures below freezing will result in a rapid temperature drop during falling precipitation.
  • Surface temperatures between 25°F and 29°F. Empirical observations have revealed that the highest-impact winter precipitation events occur within a surface air temperature range of 25°F to 29°F. Some of this is simply due to most winter precipitation events occuring on the immediate cold side of a baroclinic system, where air colder than 25°F remains farther to the north and/or west of the primary baroclinic zone. However, it has been observed that intermittent melting and refreezing of fallen precipitation results in the formation of a more slick surface of solid ice. For this to occur, the surface temperature must be both below freezing AND close enough to the freezing mark that automotive traffic can cause temporary melting of the fallen hydrometeors.

Fig 1. Refreezing of fallen sleet into solid ice due to traffic-induced melting.

  • 6 to 12 hour prior history of temperatures below 32°F. A longer period of subfreezing temperatures prior to the onset of precipitation allows road surfaces to reach the freezing mark.
  • Precipitation type of freezing rain or freezing drizzle. The slickest road surface conditions are created by freezing rain and freezing drizzle, and these precipitation types result in a hazardous road surface lacking in visual cues for motorists.
  • Precipitation amounts. Any QPF correlating with subfreezing surface temperatures indicates a risk of hazardous road conditions. Higher hourly QFP amounts correlate with a more rapid onset and maintenance of hazardous conditions. Frozen hydrometeors also have a cooling effect on surface temperatures.
  • A region's de-icing capacity. Southern US states have very little de-icing capability (salt/brine stockpiles, salt/brine trucks, plows and crews) to treat and clear roads during winter precipitation events.
  • Unanticipated winter precipitation. An unexpected winter precipitation event can cause hazardous road conditions before deicing efforts can be fully deployed. Conditions where this factor would be indicated would be short-term forecasts for winter precipitation OR developing/observed winter precipitation in areas where there have not been winter weather advisories, watches or warnings. Even in regions with adequate de-icing capability, crews and trucks take at least 2 hours to mobilize for an event.

Proposed model/mesoanalysis parameter

Let's turn the above information into a series of if-then statements, that when added cumulatively, result in a LCR value that can be output into gridpoints in forecast models and realtime mesoanalysis displays. We'll designate the parameter LCR (vehicle Loss-of-Control Risk). LCR starts with a default value of 0 and is intended to be a short-term (1-hour) parameter.

  • LCR activation criteria: LCR remains at 0 until one of the following criteria for a gridpoint is met:
    • Snow or sleet (any QPF) occuring at temperatures at or below 36°F
    • Rain or drizzle (any QPF) occuring when temperatures are at or below 32°F, OR if surface temperatures are below 36°F AND surface wet bulb temperatures are below 32°F.
    • Surface temperatures (T) at or below 32°F AND relative humidity (RH) >=98% (freezing fog potential).
  • Precipitation amounts (QPF). When:
    • Surface temperatures are below 36°F AND
    • Surface wet bulb temperatures are below 32°F
    and 1-hour liquid or liquid-equivalent QPF amounts are:
    • >=trace and <0.1", then LCR + 1.
    • >=0.1" and <0.25", then LCR + 2.
    • >=0.25", then LCR + 3.
  • Surface temperatures (T) at or below 32°F. If T <= 32°F, then LCR + 1.
  • Surface temperatures (T) between 25°F and 29°F. If T <=29°F AND T >=25°F, then LCR + 2.
  • 6 hour prior history (Th) of temperatures below 32°F. If the 6-hour average temperature T has been <=32°F, then LCR + 1.
  • Precipitation type (Ptype) of freezing rain or freezing drizzle. If Ptype is freezing rain or freezing drizzle, then LCR + 2.
  • A region's de-icing capabilities (D). A geographical factor (D). In regions with robust de-icing capability, no change to LCR. If occuring in regions with little to no deicing capability, then LCR + 3. The current methodology is to apply this factor for regions south of the 35° latitude.

Fig 2. Approximate geographic area of reduced de-icing capacity in the USA.

  • Unanticipated winter precipitation (U)*. If winter precipitation is occuring with no prior winter weather advisory/watch/warning issuance within the past 3 hours, then LCR + 2.
  • If QPF=0 AND surface temperatures (T) are at or below 32°F AND relative humidity (RH) >=98% (freezing fog potential). If QPF=0 AND T <= 32°F AND RH >=98%, then LCR = 3. Freezing fog LCR is a standalone value that should not be invoked if QPF>0. If QPF>0, freezing fog potential does not influence LCR.
Maximum LCR value The maximum value for LCR using the above criteria is 14.

LCR color scale and risk levels

The proposed LCR color scale mimics the SPC convective severe weather and WPC flood risk outlook scales.

Fig 3. LCR color/risk scale.

Example LCR values for selected events

  • February 11, 2021 - Fort Worth, Texas: LCR = 11
  • December 16, 2016 - St. Louis, MO: LCR = 10
  • February 15, 2021 - Houston, Texas: LCR = 10
  • January 6, 2017 - Birmingham, AL: LCR = 10
  • January 18, 2015 - New Brunswick, NJ: LCR = 9
  • November 1, 2021 - North Platte, NE: LCR = 4


LCR is a short-term parameter intended to display the realtime and future forecast risk of winter precipitation road impacts. It does not account for lingering impacts from untreated roads after precipitation has fallen. It can be assumed that if surface temperatures remain below freezing and no road treatment has taken place, then the vehicle loss-of-control risk will remain at its previous maximum level until either salt/brine treatment is applied or surface temperatures warm above freezing.

Current development status

November 14, 2021: The following is the current state of the Python script that will be run on each model gridpoint to calculate the LCR value for that location.

Items that need fixed or developed:

  • .to functions returning an error: the Metpy Celsius to Fahrenheit conversion functions produce the following error: AttributeError: 'DataArray' object has no attribute 'to'
  • Cannot perform function with conditionals in a vectorized manner: lcr_calc function returns error: ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
  • To be developed: function to create 6-hour temperature history

Current Python script

import pygrib
import numpy as np
import xarray as xr
import cartopy.crs as ccrs
import metpy.calc as mpcalc
from metpy.units import units
import metpy

ds = xr.open_dataset("nam.grb2")

surfacetempc = ((ds["t2m"] -273))

surfacerh = ds["r2"]

qpf = (ds["tp"] / 25.4)

surfacedpc = mpcalc.dewpoint_from_relative_humidity(ds["t2m"], ds["r2"])

surfacedp = surfacedpc.to('degF')

surfacewbc = mpcalc.wet_bulb_temperature(ds["prmsl"], ds["t2m"], surfacedpc)

surfacewb = surfacewbc.to('degF')

surfacetemp = surfacetempc.to('degF')

def lcr_calc():

lcr = 0

if (qpf > 0 and qpf < 0.1 and surfacetemp <= 36 and surfacewb < 32):

lcr = 1

if (qpf >= 0.1 and qpf < 0.25 and surfacetemp <= 36 and surfacewb < 32):
lcr = 2

if (qpf >= 0.25 and surfacetemp <= 36 and surfacewb < 32):
lcr = 3

if (lcr != 0 and surfacetemp <= 32):
lcr = lcr + 1

if (lcr != 0 and surfacetemp >= 25 and surfacetemp <= 29):
lcr = lcr + 2

if (lcr != 0 and surfacetemphist < 32):
lcr = lcr + 1

if (lcr != 0 and fzqpf != 0):
lcr = lcr + 2

if (lcr != 0 and latitude <= 35):
lcr = lcr + 3

if (qpf == 0 and surfacerh > 99 and surfacetemp < 31):
lcr = 3

if (qpf == 0 and surfacerh > 98 and surfacerh <= 99 and surfacetemp < 31):
lcr = 2

if (qpf == 0 and surfacerh > 97 and surfacerh <= 98 and surfacetemp < 31):
lcr = 1

if (qpf == 0 and surfacerh > 97 and surfacetemp < 30):
lcr = 2

if (qpf == 0 and surfacerh > 97 and surfacetemp < 28):
lcr = 3

if (surfacetemp > 36 and surfacewb > 32 and surfacetemphist > 30):
lcr = 0

if (surfacetemp > 38 and surfacewb > 32):
lcr = 0

if (qpf == 0 and surfacerh <= 97 and surfacetemp > 31):
lcr = 0

Do you have experience working with grib files and model data visualizations? Your help would be appreciated! Contact me here.

The following comments were posted before this site switched to a new comment system on August 27, 2016:

Recent Blog Posts
- Developing a road impact parameter for forecast models and mesoanalysis - October 29, 2021
- The deadly Fort Worth, Texas pileup of February 11, 2021: Its cause, contributing factors and future prevention - March 31, 2021
- Major road icing event affects 41 states - January 14-18, 2018 - January 20, 2018
- Trip to Mississippi and Alabama for Southern US winter storm - January 5-7, 2017 - January 22, 2017
- High-impact freezing rain/drizzle road icing event from Oklahoma to the East Coast - December 20, 2016
- New educational winter driving videos released, more planned - November 21, 2016
- "Icezilla" freezing rain disaster in the northeast US - January 19, 2015
- December 30-January 2 Oklahoma/Texas icing event sequence - January 2, 2015
- Freezing rain disaster - November 22, 2014 - November 23, 2014
- Report: Major winter storm in the Deep South, 1/28 - January 30, 2014
- Report: High Risk road icing event in south Texas/Louisiana - January 26, 2014
- Winter tires: not the solution for preventing icy crashes - December 13, 2013
- Preliminary low estimate of December 3-8 toll - December 9, 2013
- The Top 7 Icy Road Myths - December 6, 2013
- Aren't icy road crashes caused by driving too fast for conditions, not ice and snow? - November 14, 2012
- Road icing news for January-February 2012 - February 25, 2012
- Commentary on the recent viral WV pileup video - January 8, 2012
- Significant road icing outbreak: Monday, January 2, 2012 - January 2, 2012
- No vehicle can safely go highway speeds on icy roads - December 23, 2011
- Another case for "Road Ice Warnings": December 8-9, 2011 light snow event - December 10, 2011
- A big thanks to those helping raise road icing awareness - November 18, 2011
- 15 road icing deaths in the past week: estimating the full impact - November 4, 2011
- 2011-2012 road ice season in the US already in full swing - October 22, 2011
- Coming this winter: another 2011 tornado season toll - July 21, 2011

Past Blog Archives
- April 2010 - January 2011 posts
- December 2009 - March 2010 posts
- November 2009 posts
- October 2009 posts
- March-April 2009 posts
- February 2009 posts
- January 2009 posts
- December 2008 posts

Educational Winter Driving Videos - Watch for Free:

Video: How to correct a slide on an icy road (and how to prevent them)Video: Icy Bridges: Weather's underrated killerVideo: Deadliest Weather: Freezing Rain

Home | Risk | Stats | Accident Video | Icy Bridges | Tips | Warning Signs | If You Slide | If You Wreck

Average annual
deaths in the US
from icy roads

Average annual
injuries in the US
from icy roads

[ More Statistics ]

All content © Dan Robinson. All usage, with limited exception, requires a paid license - please see usage options or contact Dan for inquiries.

Web Site Design and Internet Marketing by CIS Internet