ABSTRACT
Risk caused by low temperatures and frost events can be very significant and may have severe impacts, even in countries with mild climate, such as Greece. Safe and timely transportation during winter months is greatly affected by the prevalence of extreme weather conditions.
In this study, temperature fluctuation and frost risk have been assessed on a mountainous major motorway in northern Greece with the use of Earth observation data and thermal mapping. More specifically, a 26 km portion of Egnatia Motorway was studied for the period January 25–31, 2008, during which frost conditions prevailed throughout the selected stretch of road.
In order to predict frost hazard levels, a Geographic Information System model was employed, which incorporates geographical and environmental factors. Two different parameterizations were applied using weighted linear combination (WLC) and analytical hierarchy (AH). Risk areas were detected and analyzed for the selected portion of the motorway. The results show that about 44%–46% of the studied length of the motorway is assessed to be in moderate to very high frost hazard levels, based on WLC and AH frost models, respectively.
Thermography surveys conducted during the period of study enabled us to better understand the temperature patterns and processes that affect its variation. The fluctuations of road surface temperature (RST) and air temperature (Tair) were produced by processing thermal mapping measurements. In addition, land surface temperature (LST) variation was produced by remotely sensed data by Moderate Resolution Imaging Spectroradiometer (MODIS), and the results were compared with the temperature profiles produced by thermal mapping. Generally, similar trends were observed in terms of temperature trends recorded by thermal mapping and MODIS data. Cold spots exhibited a consistency, as the certain locations had the tendency to remain colder.
The results of the GIS model were evaluated in comparison to temperature fluctuation, as it was produced by thermal mapping and MODIS imagery, and they were in good agreement. High Pearson’s correlation coefficient was reported between the results of the AH GIS model and the RST (r: 0.56) and air temperature variation (r: 0.51) produced by the thermography and marginally high correlation with LST variation (r: 0.49). The WLC GIS model demonstrated moderate to marginally high correlation with RST, Tair, and LST variation, with r values 0.47, 0.64, and 0.39 respectively.
This study has demonstrated the application of a GIS frost risk model on a motorway for the prediction of frost, as well as the exploitation of a range of sources that provide information on temperature along a motorway. These tools can provide us with very important information on temperature variation trends and have the potential to be used during technical project studies as well as for the planning of prevention measures against frost conditions on roads.
