HS4Cast - a reliable and automatic ice warning and
weather forecast system

The goal to minimise the risk of icy roads can only be accomplished by looking into the future. It is obviously not enough to rely on an ice sensor which only reflects the current state of the road. To give road maintenance a fair chance to take countermeasures there must be an alert at least three hours in advance. This can be done by using a model that takes all influences on the road surface temperature into consideration. Moreover, it is necessary to be informed of the state of the road, whether it is/will be dry or wet. We have developed a computer model which is able to forecast the road temperature up to 24 hours in advance using local measurements only. The typical accuracy for a three-hour prediction during a whole winter season is better than 0.75 degrees and is lower still during night-time.

Typical numbers for the forcast are:

These data were computed using a 6 month statistic with 38000 values. No intervention of meteorologists were carried out!

The table shows numbers for the median absolute errors for predictions of the road surface and the air temperatures, respectively. We use the median absolute error as a measurement of the error of the prediction because in the range +/- median absolute error are 50% of the forecasts.

To compare these results we took the 24-hour forecasts for the air temperature at noon in Vienna for the winter 1992/93. The forecasts were published in a daily newspaper and were composed by the Vienna Met Office. The median absolute error of these forecasts is 2.5 °C.

HS4Cast is available in two flavours

1. Central forecasts on a PC for one or more  measurement sites

2. Local computation of forecasts at the measurement site (using a microcomputer)

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.

Advantages of HS4Cast

	Forecast of the road surface temperature using forecasts of all influencing parameters
	Statistics open for the end user
	Automatic prediction of the road state without manual intervention, no additional manpower needed
	Optional use of large-scale data coming from a weather center; useful for the prediction of precipitation
	Observation of weather patterns using fuzzy logic enables the use of different forecast methods
	Automatic prediction of condensation and/or the buildup of ice on the road surface without human intervention
	Self-learning forecast model; this enables the reaction to changes in the horizon (new trees, buildings etc.) or the surroundings (e.g. reflecting snow fields)
	Anticipated forecast error displayed graphically
	Comparison of older forecasts with corresponding measurements possible
	Unnecessary de-icing can be avoided, thus better engagement of labour and equipment
	Enhanced road safety: accidents will be avoided
	Ecological due to careful use of de-icing chemicals
	Reduction of damages on bridges due to reduced usage of salt

Other applications

	Fruit-growing, viniculture
	Use of snow-cannons for skiing events
	Energy-saving control of heating and cooling
	Avalanche warning systems
	Detailed weather information for tourists
	Evaporation control for hydro-electric power stations
	Melting of glaciers

Security in every weather situation ... HS4Cast.