|
|
|
|
HS4Cast - a reliable and automatic ice warningThe automatic forecast system HS4Cast Version 2 computes a forecast of the road temperature to produce reliable ice warnings. For example, a three-hour forecast of the road temperature is carried out with an average accuracy of 0.75 °C. HS4Cast has been developed by the Austrian engineer Dr. G. Schaffar. Constant refinement has lead to a world-class autonomous forecast system. New technology like fuzzy logic is used to reach extreme forecast precision. Since HS4Cast not only predicts road temperatures but also many meteorological parameters (which have essential influence on the road temperature), it also can be used in applications different from ice warning on roads.
Forecast of meteorological parametersDuring critical weather situations the road condition may vary significantly within a few hundred yards. Such dramatic changes are not only influenced by varying landscape, for instance by trees, buildings, mountains or height differences; also, changing road material or bridges may have a significant impact on the condition of the road. Weather centers provide forecasts on a larger scale showing the general trends. Nevertheless, road maintenance is interested in high-precision forecasts for a relatively small area, especially regarding the prediction of the road surface temperature. To solve this task HS4Cast has been developed as a self-sufficient ice warning system. HS4Cast provides a three-hour forecast of - for instance the road surface temperature - with a mean error of less than 1 °C. Typical numbers are:
road temperature forecast air temperature forecast
mean errors in °C mean errors in °C
forecast for night 24h forecast for night 24h
1.00h 0.184 0.285 1.00h 0.293 0.346
2.00h 0.360 0.521 2.00h 0.459 0.582
3.00h 0.522 0.748 3.00h 0.599 0.770
4.00h 0.679 0.931 4.00h 0.738 0.939
5.00h 0.844 1.132 5.00h 0.895 1.088
6.00h 1.006 1.219 6.00h 1.072 1.205
10.00h 1.593 1.574 10.00h 1.566 1.450
24.00h 1.881 1.733 24.00h 2.084 1.960
The computer model HS4Cast must be plugged between a roadside sensor station and a display which monitors the measurements and forecasts. It is irrelevant where the forecasting is done: it may be carried out at the sensor station or at a central building. In the latter case more than one sensor station could feed data into HS4Cast yielding forecasts for more than one site. The forecast model
The weather data from a roadside station are classified according to the current "type of weather". Using this classification the correct mix of thermal models, statistical models and extrapolations is used to compute the forecasts for the desired intervals. HS4Cast learns the specific characteristics of a measurement site and adjusts its internal parameters individually. This is done primarily during the first days of operation and does not need human intervention. Advantages of HS4Cast
HS4Cast is available in two flavours
The central computation of forecasts is carried out by a PC under Windows NT or Windows 2000. This type of operating system enables safe parallel processing of many tasks, for example archiving meteorological data from many sites, computation of forecasts and visualization of alarms for these sites. The second variant of HS4Cast comes on a dedicated microprocessor board which is installed at the measurement site in combination with the A/D conversion and data transmission equipment. Thus, a sensor site will be transformed into a "smart" station which not only provides measurement data but also forecasts of these data. Both measurements and forecasts can be transmitted to a central station. The dedicated forecast module does not require a particular transmission protocol. Other applications
|
Send E-Mail with questions or comments regarding this Web Site to the Webmaster. Copyright © 1997 - 2004 Forschungsinstitut für technische Physik; A-2081 Hofern 14; Tel. +43 2949 7060; Ein Betrieb im Retzer Land; Last modification: 2003-02-05 |
To reach the forecast
precision mentioned, investigations of the thermal behaviour of roads as well as several
thousand computer simulations were necessary. Our computations showed that every
measurement site must be treated individually. Moreover, every weather pattern as well as
every prediction interval need a specific forecast method. For example, a forecast for 30
minutes uses a method different from a forecast for three hours. 
