The appearance of ground frost is of vital importance in construction and maintenance of roads in cold climates. Frost often causes ground heave and subsequent road damage, which must be taken into account in designing the road structure. Frost depth, pavement temperature, and freezing/thawing cycles are also important for estimating the frequency of road maintenance and treatment. Various analytical, numerical, and empirical models have been developed to estimate the surface temperature of the pavement and to model the heat flow in the underlying layers. The pavement surface experiences a variety of intricate nonlinear heat transfer mechanisms during winter, making it challenging to accurately model the surface boundary. Dynamic variation of parameters such as cloud cover and traffic density during the modeling period introduces additional complexity. To address this challenge, we have established an experimental setup in Luleå, Sweden, to measure pavement profile temperatures during the winter season. Additionally, we have developed a Finite Difference Model that utilizes local weather data including dynamic cloud cover, and which also takes traffic into account. The experimental and simulation findings demonstrate how the impact of surface temperature fluctuations diminishes and, more or less, vanishes for depths more than 55 [cm] below the pavement surface. The Finite Difference Model presented in this study exhibits the ability to forecast the pavement profile temperatures, including the surface temperature based on weather conditions, with acceptable precision for at least 3 days. As a consequence, a reasonable assessment of pavement layer conditions appears feasible based on local weather conditions, and the model can serve as a useful tool for planning road maintenance and construction in cold regions.