Extreme temperatures in urban environments exacerbate thermal discomfort and intensify the Urban Heat Island (UHI) effect, particularly during peak warm periods. Pavements, which constitute a significant portion of urban surfaces, contribute significantly to heat retention, whereas soil and vegetated areas aid in cooling through lower heat storage and higher moisture retention. Accurate forecasting of soil and pavement surface temperatures is critical for developing effective UHI mitigation strategies. This paper explores the application of Resistance-Capacitance (RC) models, a type of grey-box model, for soil surface temperature prediction. Unlike purely physics-based and data-driven models, RC models integrate physical principles with data-driven insights, balancing accuracy and interpretability. The proposed methodology is validated using real-world data from a dike in the Netherlands, where an optimal RC model is identified through an iterative process based on the Akaike Information Criterion (AIC). Results demonstrate that a two-node RC model provides a reliable balance between complexity and predictive accuracy, achieving an R² of 0.862 and a mean absolute error (MAE) of 0.675°C. These findings highlight the feasibility of applying RC models for soil temperature prediction while maintaining physical interpretability. Future research could extend this methodology to various soil types and urban surfaces, including pavements, to further enhance predictive capabilities and inform climate-responsive urban design.