Soil texture plays a critical role in determining soil properties, which are essential for plant growth and land use practices. This study aimed to spatially model soil particle size distribution (sand, silt and clay) and soil texture, using two machine learning models of Random Forest (RF) and Random Forest combined with Co-Kriging (RF-CoK). The research was conducted in the Mashhad Plain, Khorasan Razavi Province in northeastern Iran, using high-resolution auxiliary data derived from digital elevation model (DEM) and satellite imagery. A total of 180 surface soil samples (0-10 cm) were collected using a conditioned regular grid sampling method (similar to 3 x 3 km). A wide range of easily accessible environmental covariates, including remote sensing (RS) indices and topographic parameters, were considered as predictors. The most relevant variables were selected using the variance inflation factor (VIF) and the Boruta feature selection algorithm. Results indicated that the RF-CoK model outperformed the RF model in predicting PSD. For sand, silt and clay in the validation dataset, RF-CoK achieved coefficient of determination (R-2) values of 0.74, 0.65 and 0.62; root mean square error (RMSE) values of 7.12%, 8.37% and 2.85%; and mean absolute error (MAE) values of 5.70%, 6.70% and 2.28%, respectively. The Multi-resolution Ridge Top Flatness Index (MrRTF) was identified as the most influential environmental variable contributing to the spatial distribution of soil texture components. The generated soil texture maps provide valuable information for soil management and environmental monitoring. This approach, which relies on accessible and cost-effective data sources, offers strong potential for application in other agricultural regions with similar conditions.