MDSS Lab Prototype: Road Weather Forecast System Enhancements

MDSS Lab Prototype: MDSS Lab Prototype: Road Weather Forecast System EnhancementsRoad Weather Forecast System Enhancements

Bill Myers

National Center For Atmospheric Research(NCAR)

MDSS Stakeholder MeetingBoulder, CO20 October 2005Photo by Dave Parsons

2

Overview

• RWFS Enhancements through Release 4

• Frost Deposition Module• Ongoing Upgrades

3

RWFS Enhancements in Release 4.0

• Precip variables’ weights hard-wired– Quality observations lacking– Spatial and inter-variable consistency

• Added snow-water ratio algorithm– Better snowfall estimate

• Refined insolation data sources (model blend)– Model data compared to Vaisala sensor

• MOS forecast interpolation– Improved forecasts where interpolation makes sense

4

Frost Deposition Module

• Based on ISU Frost Deposition Model (Tina Greenfield)

• Tries to capture uncertainty in forecast– Frost deposition is very sensitive– Monte Carlo approach varies air temperature, dew point,

and wind speed

• Fuzzy logic interest map applied to each permutation’s output– More credence given to larger frost accumulations– Likelihood of forecast permutation related to weight

• Output– Frost potential (not probability of frost)– Can be thresholded to provided Low, Medium, High alerts

5

Frost Module Operations

Weather Forecasts

Road Temp Forecasts

Bridge Temp Forecasts

Bridge Frost Forecasts

Road Condition Forecasts

• Road Temp

• Bridge Temp

• Bridge Frost Potential

MDSS 4.0 generates Bridge Frost forecasts though Road Frost could be calculated

6

Ongoing Weather Forecast Improvements

• Improved forecasts of extreme events– Model Error Correction (MEC) captures rare events better

than current scheme (Dynamic MOS)

• Incorporation of radar data– Precipitation extrapolation provides better forecast in first

few hours (than model/RWIS combination)– Need to trend toward model forecast seamlessly– Provides improved spatial pseudo-observations

• Higher resolution NCEP data– GRIB-2 encoded model data is available– Fully uncompressed data sets are unwieldy– Need tools to extract only relevant information