Information on pavement layer thickness is very important for determining bearing capacity, estimating remaining life and strengthening planning. Ground-penetrating radar (GPR) is a nondestructive testing (NDT) method used for determining the continuous pavement layer thickness in the travel direction. The data obtained with GPR in one survey line is suitable for the needs of repair and rehabilitation planning of roads and highways, but not for wider traffic areas such as airfield pavements. Spatial representation of pavement thickness is more useful for airfield pavements but requires a 3D model. In the absence of 3D GPR, a 3D model of pavement thickness can be created by additional processing of GPR data obtained from multiple survey lines. Five 3D models of asphalt pavements were created to determine how different numbers of survey lines affect their accuracy. The distance between survey lines ranges from 1 to 5 m. The accuracy of the 3D models is determined by comparing the asphalt layer thickness on the model with the values measured on 22 cores. The results, as expected, show that the highest accuracy is achieved for the 3D model created with a distance of 1 m between survey lines, with an average relative error of up to 1.5%. The lowest accuracy was obtained for the 3D model created with a distance of 4 m between the survey lines, with an average relative error of 7.4%.

Road traffic noise is the second largest environmental stressor in urban areas in Europe after air pollution. The harmful effects of noise arise mainly from the stress response it triggers in the human body, which can have significant consequences for physical and mental health. Therefore, the use of a reliable noise prediction model is an important prerequisite for the quality assessment of a number of residents exposed to excessive noise levels and for the selection of appropriate noise mitigation measures. In this study, the analysis of road traffic noise in an urban street in the narrower centre of the Croatian capital Zagreb was performed using four noise prediction models: "RLS-90", "RLS-19", “NMPB-Routes-96 (SET-RA-CERTU-LCPC-CSTB)”, and “CNOSSOS-EU”. LimA V2021 noise prediction software was used for the analysis, and the noise modelling results were validated with short-term noise measurements. The main objective of the research presented in the article was to test the “CNOSSOS-EU” method, recently introduced in Croatian noise control practice, and to gain initial insights into which of the aforementioned noise prediction models is the most reliable for the assessment of road traffic noise in urban environments in Croatia. A comparison of the noise modelling results with the results of short-term noise measurements has shown that the German national calculation methods “RLS-90” and “RLS-19” as well as the “CNOSSOS-EU” method provide significantly more accurate noise predictions than the “NMPB-Routes-96 (SETRA-CERTU-LCPC-CSTB)” method.