Over the past two decades, numerous studies have emphasised the importance of including organic matter (OM) in land surface models (LSMs) to accurately represent soil thermal and hydrological properties. This is particularly relevant in Arctic regions, where organic-rich soils are widespread. Consequently, most LSMs incorporate parameterisations that account for OM effects, although these implementations are often simplified. Recent advancements in global soil datasets now enable more precise modelling of soil properties by providing detailed inputs for soil composition and physical characteristics. This study enhances the ORCHIDEE LSM by refining the representation of soil organic and mineral content, as well as improving parameterisations of heat capacity, thermal conductivity, and porosity, using data from the SoilGrids 250m v2.0 database. The updated model is evaluated across multiple Arctic sites and compared against two earlier versions: (1) a Bulk version that neglects OM effects and (2) a simplified version with a basic OM prescription. Results show that incorporating OM into thermal process modelling significantly improves soil temperature simulations, particularly at greater depths. For some sites, root mean square errors (RMSE) are reduced by up to 25% compared to the Bulk version, especially during the snow-free summer months. These findings highlight the value of high-resolution soil datasets, such as SoilGrids, for improving simulations of thermal dynamics in carbon-rich Arctic soils.