LCR (vehicle Loss-of-Control Risk): A new winter weather parameter and scale (v1.2.2)

1. Latest Forecast Chart
2. Video Presentation
3. Purpose
4. Justification
5. Applications
6. LCR Scale and approximate human impacts
7. Methodology
8. Summary
9. Color scale values
10. BFP and BFP+, Below-Freezing Precipitation
11. Script Downloads & Development
12. Script Version Notes
13. Running the LCR script on model data
14. Current LCR & BFP Forecast Charts
15. Development Goals
16. Limitations
17. Known issues
18. GNU License Info
19. Collaboration
20. Special Thanks To

• Number of accidents
• Number of injuries
• Number of fatalities
• Demand on police/fire/EMS first responders
• Road closures/restrictions
• Property damage losses
• Demand on towing services
• Shipping and commerce impacts

A way to visualize this condition has not previously existed in publicly-available forecast tools. For example, discerning the simple condition where precipitation is falling at below-freezing surface temperatures has involved toggling back and forth between QPF (precipitation) and 2m temperature charts:

ABOVE: Icy roads are a risk where precipitation overlaps with subfreezing surface temperatures. The forecaster must toggle back and forth between QPF and 2-meter surface temperature charts to attempt to visualize where that overlap is occurring.

Even snowfall charts do not provide this visualization, as many light snow events have large areas where the precipitation is falling into above-freezing temperatures:

ABOVE: Snow accumulation or snow precipitation type charts do not show the QPF/freezing temperature overlap. Snow often falls into surface temperatures both above and below freezing, and these conditions are not revealed by the snow accumulation charts alone.

The BFP (below-freezing precipitation) chart, the precursor to LCR, was developed to show the precip-freezing temperature overlap:

ABOVE: Simple BFP (below-freezing precipitation) chart for the same time as the examples above.

The BFP chart simply showed only the QPF that was occurring at temperatures at or below freezing, visualizing the overlap of these two fundamental ingredients. BFP+ charts show all QPF with three additional temperature categories, each with thier own color scales:

ABOVE: BFP+ chart for the same time as the examples above.

LCR expands on the basic BFP chart, consolidating all of the meteorological factors that produce and increase the hazardous road condition risk into a single, easy-to-understand parameter:

ABOVE: LCR chart for the same time as the previous examples.

It is intended for LCR to be a complement, not a replacement, for other winter weather hazard assessment tools. The Winter Storm Severity Index, for example, is a product that consolidates ALL winter weather hazards such as extreme temperatures, travel impacts, property damage and power outages. LCR is a product that is focused solely on the danger to drivers losing control from a significant reduction in road surface friction.

• LCR charts can be generated from realtime meteorological data
• LCR forecast charts can be generated from numerical model data
• LCR can be easily manually calculated for any in-progress event at a given location
• LCR can be manually calculated for past events at a given location
• LCR charts for past events can be generated from meteorological data archives

Base LCR (v1.2.2)

Factors that increase LCR (v1.2.2)

LCR values for select past events

• 2m temperature at or below 38°F (if precip type is snow) 36°F (all other precip types)
• 2m wet bulb temperature at or below 36°F (if precip type is snow) 34°F (all other precip types)
• Precipitation of any type (QPF) is greater than zero

BFP Plus is an enhancement of BFP that includes three additional data sets: NFP (Near-Freezing Precipitation), AFP (Above-Freezing Precipitation) and CIP (Precipitation at/below Critical Icing Temperature). On a BFP+ chart, the BFP, NFP, AFP and CIP data sets are all plotted together to show the areas at risk of icy roads within the context of all areas of precipitation.

• CIP (Precipitation at/below Critical Icing Temperature): The amount of precipitation (QPF) of any type occurring where surface temperatures are at or below 29.9°F. This is the condition that produces the highest-impact icy road events. CIP areas are plotted on the BFP+ charts with a red gradient color scale. CIP represents a "enhanced warning" area for road icing. Drivers within CIP-shaded zones should expect hazardous road conditions, as the weather conditions are favorable for the rapid and efficient development of road icing during the time period represented on the chart.
• BFP (Below-Freezing Precipitation): The amount of precipitation (QPF) of any type occurring where surface temperatures are at or below freezing (32.9°F). This is the basic condition that produces the vast majority of high-impact icy roads. BFP areas are plotted on the BFP+ charts with a blue (turquiose-to-purple) gradient color scale. BFP represents a "warning" area for road icing. Drivers within BFP-shaded zones should be on high alert for hazardous road conditions, as the weather conditions are favorable for the development of road icing during the time period represented on the chart.
• NFP (Near-Freezing Precipitation): The amount of precipitation (QPF) of any type occurring where surface temperatures are between 32 and 38 degrees Fahrenheit. NFP areas are plotted on the BFP+ charts with an orange gradient color scale. NFP represents a "caution" or "watch" area for road icing during the time period represented on the chart. Drivers should be aware that conditions favorable for road icing could develop within NFP-shaded areas if temperatures fall more than expected and/or if the rate of sleet/snow falling is heavy enough. Essentially, NFP areas are close to, and can become, BFP areas. Drivers in mountainous terrain within NFP-shaded areas should remain vigilant for unexpected conditions with changes in elevation or when exiting tunnels.
• AFP (Above-Freezing Precipitation): The amount of precipitation (QPF) of any type occurring where surface temperatures are above 38 degrees Fahrenheit. AFP areas are plotted on the BFP+ charts with a green gradient color scale. AFP represents a "no high-impact road icing expected" area of precipitation during the time period represented on the chart. Drivers within AFP-shaded zones should exercise the normal cautions necessary for safe operation of a vehicle in rain and on wet roads. Be aware that hail can occur within AFP areas in thunderstorms. If heavy enough, hail is capable of accumulating on roads and creating a hazard even in an AFP area.

LCR v1.2.2 (Updated November 19, 2025)
The LCR scripts are designed to run in the NCO toolkit. The scripts are being released under a GNU license.

Current Version 1.2.2

Generating LCR forecasts requires these two NCO script files and at least one of the Linux shell script files in the next section.

Shell scripts for specific models: In addition to the two nco script files listed above, you will need one or more of the following Linux shell scripts to run LCR forecasts (view instructions on how to run). Use the one that corresponds to the model you wish to use to generate the LCR forecast.

When generating charts, you can import the following color table files:

Archived Older Versions of the LCR NCO script

You can also download the script from Github.

Notes

• Numerical model chart script (Phase I): The first functional LCR script for numerical model charts was released in December 2022. This code is being made available under a standard GNU license to any and all outlets that display numerical model data charts.
• Realtime analysis script (Phase II): Realtime LCR chart generation will be possible using METAR (surface) observations and mesonets. Development of this code is scheduled, but no date of completion has been set. Mesoanalysis charts can be generated using the numerical model data script.

1. Install the NCO toolkit and related functions in Linux. From a fresh Linux install, type and run the following commands in sequence:
   sudo apt-get update

   sudo apt-get install nco

   sudo apt-get install netcdf-bin

   sudo apt-get install cdo

2. Download the LCR script to your Linux installation using the links above or with wget commands.
3. Download model data and run the scripts: There are two ways to do this: using shell (bash) scripts or using manual commands.

   Using bash scripts: The LCR v1.2 script is designed to run as part of a series of commands performed in sequence with the Linux shell (bash) scripts (download links above or use the wget command). These shell scripts automate the downloading of the model data, execute the LCR script and create the Day 1 and Day 2 data files. Follow these instructions to run the bash scripts:

   • Download each .sh file into Linux (links above or using the wget command). These need to be downloaded into the same Linux folder as the LCR and Kuchera nco script files.
   • chmod each .sh file to 755 by typing the following command and pressing enter:
     chmod 755 scriptname.sh
   • Edit the date and model run time in each script. The date format is YYYYMMDD and the model run times are HHz.
   • Run the bash script by typing the following command and pressing enter:
     ./scriptname.sh
   Manual steps: Please note that while LCR v1.2 can be run manually via individual commands, there are numerous prerequisite steps that are better performed within the shell (bash) scripts. Follow these steps to do the process manually:
   • Download model data with the ncks command. The following two commands run in sequence will download the necessary parameters in model output from the NOAA Nomads server:

     ncks -v tmp2m,dpt2m,rh2m,apcpsfc,
     cfrzrsfc,csnowsfc,tcdcclm,gustsfc OpenDap-Model-URL modeldata.nc

     If doing the Kuchera ratio caluclations, you will need to next download the required upper air data with this command:

     ncks -d lev,500.0,1000.0 -v tmpprs OpenDap-Model-URL outputfile.nc time

     OpenDap-Model-URL is the location of the OpenDAP data file on the NOMADS server. This will be different for each model. Go to the NOMADS home page and click the OpenDAP link for the model you need.

     Constrain data by latitude/longitude: You may want to limit the download of data to a specific region like the CONUS or an even smaller geographic area. For example, the full global GFS data download for the LCR-required variables will be over 2GB. To save time and limit the data download to CONUS gridpoints only, use the following:

     ncks -d lat,20.0,60.0 -d lon,210.0,310.0 -v tmp2m,dpt2m,rh2m,apcpsfc,
     cfrzrsfc,csnowsfc,tcdcclm,gustsfc OpenDap-Model-URL outputfile.nc

   • If you are using the Kuchera snow ratio calculations, next run the following command to calculate Kuchera ratio from the downloaded upper air data (make sure the resulting file is named kuchera.nc):

     ncap2 -4 -S kuchera.nco downloaded-upper-air-data-file.nc kuchera.nc

     downloaded-upper-air-data-file.nc is the outputfile.nc from the upper air download step.

   • If you are using the Kuchera snow ratio calculations, next merge the kuchera.nc file with the rest of the downloaded model data parameters (modeldata.nc shown above, you can name it whatever you want):

     cdo merge modeldatafile.nc kuchera.nc modeldatafile-merged.nc

   • Run the LCR script on the downloaded model data file. Run the following command to calculate LCR and BFP from the downloaded data.

     ncap2 -4 -S lcr-v-1-2.nco downloaded-model-data-file.nc outputfile-lcr.nc

     downloaded-model-data-file.nc is the modeldata.nc from the previous download step (use the merged file if using Kuchera snow ratios), outputfile-lcr.nc here is the final file containing the LCR and BFP calculations.

   • Generate 24-hour maximum charts: The ncwa command can generate a new .nc file containing maximum LCR or BFP values over a specified time period.

     ncwa -v lcr -d time,startframe,endframe -a time -y max inputfile-lcr.nc outputfile-lcrmax.nc

     Replace startframe with the time step number where you want the maximum value calculation to begin, replace endframe with the time step number where you want it to end.

     For 1-hour models like the HRRR, there are 24 timesteps in a day. For models like the GFS which use 3-hour time steps, there are 8 timesteps in a day. If you wanted to perform a maximum calculation for the 00z HRRR for the 24-hour period ending at 00z the following day, the startframe would be 1 and the endframe would be 24. For the same situation with the 00z GFS, the startframe would be 1 and the endframe would be 8. If you wanted to do a "Day 2" forecast for tomorrow (00z tonight through 00z tomorrow) from the 18z HRRR, you would make the startframe 6 and the endframe 30.

     Example: 24-hour maximum LCR chart from the 00z HRRR:

     ncwa -v lcr -d time,1,24 -a time -y max inputfile-lcr.nc outputfile-lcrmax.nc

     Example: Day 2 24-hour maximum BFP chart from the 18z HRRR:

     ncwa -v bfp -d time,6,30 -a time -y max inputfile-bfp.nc outputfile-bfpmax.nc

4. Copy the final LCR data files to Windows (in WSL): Type the following command to open the Linux directory in a Windows File Explorer window (note the space and period after exe):
   explorer.exe .

   You can drag-and-drop or copy the .nc data files over to Windows this way. This folder also does not auto-refresh, you'll need to manually refresh it to see new/changed Linux files. Note: This file transfer menthod only works in one direction, Linux > Windows. Files copied from Windows to Linux this way are often not readable in Linux.

5. Use the data viewer of your choice to generate charts from the resulting files. The free software McIDAS-V is a good way to generate charts of the resulting data. Choose the "color shaded plan view" and download the XML files to import the color scale for LCR or BFP, and set the scale range from 0 to 12 for LCR or 0 to 0.1 for BFP.
   

• Refined QPF threshold values for base LCR assignments.
• Lowered the main threshold temperature of liquid precipitation to 36°F and wet-bulb temperature of 34°F. (Snow threshold remains at 38°F and wet-bulb temperature of 36°F)
• Added lower bounds for all base LCR value assignments (see tables above).
• Added lower bound of 0.003" for all BFP+ variables.
• Added function to turn off QPF decumulation for the HRRR model, which already has hourly values at each timestep.
• Added a LCR limit of 7 when precip type is snow outside of the southern USA (north of 35° latitude) and wind gusts are less than 20mph. The reason this was done is to preserve values 9 and above for only freezing rain/drizzle events in the central and northern USA. LCR-7 events will be asssigned LCR-8 only when high winds are present during those snow events.

• For models without asnowsfc, the Kuchera ratio is now used to calculate the one-hour snowfall amounts used to set base LCR. Because calculating Kuchera ratio requires a significant increase in model data variables, the required NOMADS OpenDap data download size for each run has increased. The included companion script kuchera.nco handles this portion of the calculation.
• The precipitation criteria for base LCR have been lowered. base LCR is now set as follows if precip type is snow: 0" to 0.1": LCR-1; greater than 0.1" to 0.6": LCR-2; greater than 0.6" to 1.3": LCR-3; greater than 1.3": LCR-4. base LCR is now set as follows if precip type is liquid: 0" to 0.05": LCR-1; greater than 0.05" to 0.1": LCR-2; greater than 0.1" to 0.25": LCR-3; greater than 0.25": LCR-4. These changes were implemented to better match impacts observed.
• A single script now processes data from any model, including both asnowsfc and non-asnowsfc inclusive models.
• The lower bound for the temperature "sweet spot" factor has been lowered from 24°F to 20°F, increasing the range of temperatures for which this factor will trigger.
• The at-or-below freezing threshold has been increased to 32.9°F in order to ensure triggering of the factor when temperatures are right at the freezing point.
• The script now uses the variable tcdcclm (total cloud cover percentage) to limit LCR to 3 when skies are clear (cloud cover percentages less than 10). This was done because models frequently depict clear-sky freezing precipitation during frost conditions. These conditions very rarely produce impactful icing, yet were frequently resulting in LCR values of 7 being depicted on charts.
• Added a new surface wind LCR incremental factor that looks at the model variable gustsfc (maximum surface wind gust). While the windspeed thresholds will likely require some future adjustments, they are being initially set to the following: If maximum surface wind gusts are greater than 8.9 m/s (17.4kts or 20mph) and LCR is at 5 or above, LCR is increased by 1. Justification: Strong winds (crosswinds in particular) increase the risk of losing control on icy roads, particularly for high-profile vehicles like tractor-trailers, vans and RVs. Wind alone can cause lateral slides/fishtails for higher profile vehicles without any driver steering input.
• Scale labels have been simplified. LCR 1 through 3 are labeled as Chance, LCR 4-5 are labeled as Likely and LCR from 6 and up are labeled as Serious. The scale's colors remain the same.
• Color scale updates: Updated the BFP color scale to a better turqiouse-blue-purple gradient. Added GEMPAK TBL format downloads for all color scales.
• Added BFP+ chart functionality: Version 1.2 of LCR creates the new BFP Plus data set. BFP+ is a bundle of four parameters, each with their own period maximum .nc file: CIP (Precipitation at/below Critical Icing Temperature), BFP (Below-Freezing Precipitation), NFP (Near-Freezing Precipitation) and AFP (Above-Freezing Precipitation). BFP-only charts can be plotted standalone in the same way previous versions. BFP+ charts are plotted by overlaying AFP (bottom layer), NFP (third layer), BFP (second layer) and CIP (top layer) using your data visualization ot mapping software of choice such as McIdas. Make sure you download the new BFP+ color scale files and apply each to their respective data set.

• Corrected a significant error with the base LCR assignment code. The LCR-3 base condition was actually assigning a value of 2 instead of 3. v1.1.7 has the same correction for models without asnowsfc.
• v1.1.7 only (models without asnowsfc): Added a function to set base LCR based on snow amounts using the 10:1 ratio. This is another measure to help ensure that heavy snow events will attain more appropriate LCR values. It assigns LCR values from 1 to 4 based on 10:1 ratio snow amounts if the precipitation type is snow, using the same thresholds as models with asnowsfc. This new function requires the additional variable csnowsfc for each gridpoint. Freezing rain will remain using the liquid thresholds.

• Increased the wet bulb temperature threshold for base LCR by 1 degree: At or below 34°F for LCR-1 and at or below 33°F for LCR-2 through LCR-4. This was done so that a greater number of heavy snow events in above-freezing 2m temperatures can attain values up to LCR-4. v1.1.5 has the same modification for models without asnowsfc.

• Added a 4th base LCR value for heavy snow events. This assigns a base LCR value of 4 for events with 3 inches or greater snowfall in 1 hour (asnowsfc models only) and/or liquid-equivalent 1-hour precipitation of 0.5" or greater. This ensures that heavy snow events will attain a value of at least LCR-4, even if 2m temperatures are above freezing. v1.1.3 is for models without asnowsfc (meaning only the liquid-equivalent precipitation factor is used).

• Added snowfall thresholds to the base LCR value functions. The liquid-equivalent thresholds for base LCR remain the same, but now the script looks at the 1-hour snowfall (decumulated asnowsfc variable) and assigns LCR-1 for snowfall from greater than 0 to 1 inch, LCR-2 for snowfall between 1 and 2 inches, and LCR-3 for snowfall above 2 inches.

  This change was necessary because impactful events with high snow ratios (meaning very low liquid-equivalent precipitation) were sometimes not assigned base LCR values higher than 1. This resulted in events that should be LCR-4 or LCR-5 ending up at only LCR-3.

  Since the asnowsfc variable is not available in all models, version 1.1.4 of the script can only be used with models that include that variable (such as the HRRR). Deriving hourly snowfall in models without the asnowsfc variable will involve additional development to incorporate Kuchera snow ratio (to be completed at a future time).

• Removed the lower temperature bound of the "sweet spot" range for freezing rain events. The script will now apply the 2-point "sweet spot" increase for all temperatures at 29°F and below if the precipitation type is freezing rain. This is because there is no decrease in the hazard level of freezing rain road icing occurring at temperatures below 24°F.

• Fixed an error with the temperature history factor variable not converted to Fahrenheit. Since the threshold of this factor is 32, the value being in Celsius meant that it would always trigger the temperature history factor LCR increment. This caused LCR to be 1 point higher everywhere on the map and causing all LCR-activated gridpoints to start at 2 instead of 1.
• Rewrote the LCR processing section to add the lcron variable as the "master" control in the conditional factor "where" statements. This ensures the incremental factors don't activate at a gridpoint when LCR is zero.
• Increased the temperature threshold of LCR-1 to 38°F and a wet bulb temp of 33°F. This should help ensure that the vast majority of above-freezing heavy snow events trigger LCR at a gridpoint.
• Lowered the relative humidity thresholds for the freezing fog factors to 92 percent for LCR-1 and 98 percent for LCR-2. This should help to catch more of the lower-end freezing fog events that can initiate at lower RH levels.
• Added freezing rain precipitation type to the below-freezing temperature factor. This is meant to maintain increased LCR for freezing rain events occuring at temperatures above freezing.

• De-Icing Capacity refinements. Currently, the 35°, 34° and 33° latitudes are used as proxies for demarcating regions of adequate vs reduced de-icing capacity (salt/brine trucks, plows and crews). Future plans call for a detailed county-by-county database of de-icing capacity across the US that can be used to apply a granular GIS-based De-Icing Capacity factor in LCR calculations. This data will also allow for a multiple-step grading of de-icing capacity that will allow application of this factor in an incremental manner.
• RWIS integration. Current realtime analysis will use AWOS/ASOS observations. Later editions of the script will include RWIS (Roadway weather information systems) data, where available.
• Radar integration. The possibility of using radar reflectivity for realtime LCR calculations is being investigated. Radar may be used to augment or replace the QPF/precipitation rate factor for realtime LCR analysis.

• Temperatures remaining below freezing with no road treatment applied. As long as roads remain untreated and temperatures remain below freezing, the hazardous condition will remain at is previous LCR maximum until either of those factors has changed. It will remain possible for LCR values calculated in the time following an event to not show this ongoing hazard.
• Temperatures rising above freezing immediately following a high-LCR event. Road surfaces can remain hazardous for an interval of time even if temperatures rise above freezing immediately afterward.

LCR does not take into account rain (wet road) impacts. Rain does correlate with an increase in accident rates, but only at a marginal level compared to the increase during winter precipitation conditions. Rain, a very common condition in most locations, rarely results in the acute, widespread impacts that winter precipitation produces. The presence of precipitation during the warm season is already well-represented using radar and model QPF charts. In addition, including rain in the LCR equation would create unneccessary noise in the data. These all influenced the decision to not include a LCR factor for rain.

• False freezing precipitation areas: Many models, including the HRRR, produce broad swaths of light freezing precipitation in areas they should not, for reasons unknown. The QPF values in these areas are generally lower than 0.008", and rarely verify with observed icing or impacts. Actual freezing precipitation can produce high impacts with very low (trace) amounts, so placing a lower bound risks LCR missing high-impact light icing events (mainly freezing drizzle events). However, having no lower bound for freezing precipitation results in high LCR values being plotted across these broad areas of false values.

  Lower QPF bounds for freezing precipitation of 0.01" and 0.008" were tested for several weeks at a time. These resulted in LCR missing multiple high-impact freezing drizzle events. Currently, a compromise lower bound value of 0.003" is in place on the freezing rain/freezing drizzle factor in version 1.2.2.

  Since most of the trace/light icing events occur in areas of broad-scale lift, the tcdcclm (cloud cover) variable has shown some success in filtering out part of these false freezing precipitation areas by limiting LCR to 3 where cloud cover is at or less than 10%. This function is also active in v1.2.2.

  The false freezing precipitation issue is a model problem and not necessarily an LCR problem, and may require further workarounds to completely eliminate.

• QPF accumulation/decumulation issues: The various models each handle precip accumulation differently. Some do not accumulate QPF, while others accumulate either at set intervals or through the entire forecast period. Some have demonstrated irregular/random accumulation patterns. For example, the NAM NEST model produces accumulation in a 3-timestep interval, but occasionally will accumulate in seemingly random 2-timestep intervals with no apparent pattern. Since the script needs single-timestep QPF values to calculate LCR for that timestep, any accumulated QPF needs to be decumulated using a set formula.

  v1.2.2 assumes entire-forecast-period QPF accumulation and has a function to decumulate values by subtracting the previous timestep value from the current timestep value. The exception is the HRRR model, for which the script does not decumulate precipitation values.

  Complete model precipitation accumulation behavior profiles will need to be added to the script to ensure that single-timestep values are processed before LCR is calculated.

• Regional de-icing capacity factor: There has been feedback about the way the regional de-icing capacity is represented for the increase in risk in the southern USA. Since this factor uses latitude and LCR is an integer scale, this results in the increased values plotted in uniform lines across the map. These lines look out of place on any kind of weather map. The latitude is a good approximation of the actual increase in risk with southward extent, so the issue here isn't so much with the values themselves, just the unnatural way they appear on the final charts.

  A GIS layer with county-by-county de-icing capacity may be needed that can replace the generalized latitude factor. Unless that sort of data already exists somewhere, it's going to be quite the undertaking to compile it - and it's not apparent if the end result would be much more useful than the current method.

• This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 3.
• This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
• You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/.