LCR (vehicle Loss-of-Control Risk): A new winter weather parameter and scale (v1.1.6 Beta)

1. Sample 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 - Below-Freezing Precipitation parameter
11. Script Downloads & Development
12. Script Version Notes
13. Running the LCR script on model data
14. LCR & BFP Forecast Charts and Archives
15. Development Goals
16. Limitations
17. Collaboration
18. 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

LCR consolidates all of the meteorological factors that produce and increase the hazardous road condition risk into a single, easy-to-understand parameter.

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

Baseline LCR (v1.1.6 Beta)

Factors that increase LCR (v1.1.6 Beta)

LCR values for select past events

• 2m temperature at or below 38°F
• 2m wet bulb temperature at or below 34°F
• Precipitation of any type (QPF) is greater than zero

Below is a 24-hour maximum chart of BFP. The sensitivity of the BFP chart color scale (zero to 0.1" and above) is highly amplified due to the fact that it only takes very light amounts of precipitation to create hazardous road conditions.

LCR v1.1.6 Beta (Updated December 26, 2022)
The LCR scripts are designed to run in the NCO toolkit. The scripts are being released under a GNU license.

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

   sudo apt-get install lynx

2. Download the LCR script to your Linux installation. You can do this with Lynx with the following command:
   lynx https://icyroadsafety.com/lcr/lcr-v-1-1-6.nco

   Once the script loads in Lynx, press P to save it.

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 easiest way to generate LCR and BFP data is to use the bash scripts, which automate the downloading of the model data, execute the LCR script and create the Day 1 and Day 2 data files. Follow these instruactions to run the bash scripts:

   • Download each .sh file into Linux (links above). These need to be downloaded into the same Linux folder as the LCR nco script file.
   • 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: Follow these steps to do the process manually:
   • Download model data with the ncks command. The following command will download the necessary parameters in model output from the NOAA Nomads server:
     ncks -v tmp2m,dpt2m,rh2m,apcpsfc,cfrzrsfc OpenDap-Model-URL outputfile.nc

     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 OpenDap-Model-URL outputfile.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-1-1-5.nco downloaded-model-data-file.nc outputfile-lcr.nc

     downloaded-model-data-file.nc is the outputfile.nc from the previous download step, 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.
   

• Increased the wet bulb temperature threshold for baseline 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 baseline LCR value for heavy snow events. This assigns a baseline 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 baseline LCR value functions. The liquid-equivalent thresholds for baseline 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 baseline 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.