Road segment forecasting A tool for salt spreader control using GPS VIKING workshop
Post on 14-Jan-2016
32 Views
Preview:
DESCRIPTION
Transcript
March 15th 2007
Road segment forecasting A tool for salt spreader
control using GPSVIKING workshop
Hamburg, March 15th 2007by
Alexander Mahura, Claus Petersen and Bent SassPresented by
Torben Strunge PedersenThe Danish Meteorological Institute
March 15th 2007
Outline of the talk
• What is Road Segment Forecasting (RSF) ?
• A RSF method
• The Potential in RSF
• Quality and examples
• The future
March 15th 2007
Road station
March 15th 2007
Point forecasts (PF) from Road Condition Model (RCM) – Showing status
March 15th 2007
RSF - Showing status
March 15th 2007
Ice car – Used for thermal mapping
March 15th 2007
Thermal mapping, Halskov - Ringsted
Left and right lane
March 15th 2007
Mean thermal mapping profile
March 15th 2007
GPS equipped spreaders
• GPS has opened the opportunity for automatic control of width and asymmetry of the spreading pattern
• With reliable weather forecasts for segments of the route the automation can be taken a step further – to automatically vary the amount of salt applied along the route
March 15th 2007
New equipment for spreaders
• The price of infrared sensors determining the temperature of the road surface has come down
• Together with GPS this makes it economically possible to get thermal mapping data from spreaders in operation
• Provide data for validation
March 15th 2007
Infra red sensor mounted on spreader
March 15th 2007
Infra red sensor mounted on spreader
March 15th 2007
Test route – 2005 / 2006 season
Route VA 4 - Ribe County
March 15th 2007
Road segment forecast
• Based on PF ”near” the route forecasts for 2 km segments are obtained by interpolation
• Note this method does not use information from the infra red sensors!
March 15th 2007
End user presentation
March 15th 2007
Forecast quality
• Making a forecast is easy– But what is the quality ?
• Here we will look at quality through two measures– Mean Deviation (MD)– Mean Absolute Deviation (MAD)
March 15th 2007
Forecast quality NWP
• NWP MAD – PF for road stations
MAD: Road temperature February 2006 (about 27.000 forecasts)
1.22 1.17 1.19 1.19 1.19
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1 2 3 4 5
NWP
March 15th 2007
Forecast quality – Present system
• NWP versus RCM MAD – PF for road stations
MAD: Road temperature February 2006 (about 27.000 forecasts)
1,221,17 1,19 1,19 1,19
0,28
0,440,54
0,650,74
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1 2 3 4 5
NWP RCM
March 15th 2007
Forecast quality – Present system
• The importance of road stations
MAD: Road temperature February 2006 (about 160.000 forecasts)
1.16 1.14 1.12 1.1 1.08
0.29
0.470.59
0.670.73
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1 2 3 4 5
RCM: no Obs RCM
March 15th 2007
RSF-Potential - Hypothetical route
March 15th 2007
“RSF” potential
MAD: Road temperature February 2006 (about 160.000 forecasts)
0,690,78 0,82
0,910,97
0,38
0,58
0,710,81
0,89
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1 2 3 4 5
RSF RCM
March 15th 2007
Route VA 4 – thirty 2 km segments
March 15th 2007
Measurements from spreaders
• 2005 – 2006 season• 102 operations available (duration
from minutes to about 4 hours)• 90.459 measurements in all• Time averaging (1 min.) and allocation
to the corresponding 2 km road segment reduces the amount available for the evaluation to 5.844 measurements
March 15th 2007
Amount of observations by segment
March 15th 2007
Comparison of observations
Infra red sensor versus road station
Starup
MD -0.11 degree
MAD 0.51 degree
March 15th 2007
RSF (3 hour) versus observation I
March 15th 2007
RSF (3 hour) versus observation II
March 15th 2007
RSF (3 hour) versus observations MD
March 15th 2007
RSF (3 hour) versus observation MAD
March 15th 2007
The future
• The near future (now)– Integration of infra red sensor data in the RCM /
RSF forecasting– Increased horizontal resolution in the RCM– Include more test routes
• The not so distant future– Use of RSF for applying variable amounts of
salt along routes
• And then – eventually– Integration of RSF in spreaders for automatic
controlling of the salt amount used
March 15th 2007
Conclusion
• The technology for developing and evaluate RSF is available
• This pilot study shows that the quality of RSF is nearly comparable to the one known from state of the art PF– Road stations are essential for a good
quality of forecasts (both PF and RSF)
March 15th 2007
The future is very near (you)!
Thank you for your attention!!!
top related