Timely and accurate acquisition of licorice spatial distribution information can provide scientific data support for the healthy development and refined management of the licorice industry. Based on the Google Earth Engine (GEE) platform, using the main licorice-producing area of Dengkou County in Inner Mongolia as the study area, and utilizing Sentinel-1, Sentinel-2, and SRTM digital elevation data as data sources, a multi-source and multi-dimensional classification feature set was constructed. Six different schemes were designed to explore the combination of spectral features with different types of features and the impact of feature selection on licorice identification. The random forest classification method was used for licorice identification. The results show that the index features have the most significant contribution rate in licorice identification, followed by textural features, spectral features rank third, polarization features have a relatively lower contribution rate, and topographic features contribute the least to licorice identification. Selecting features can eliminate unnecessary ones, thereby enhancing the precision of identifying licorice, achieving an overall accuracy of 91.15% and a Kappa statistic of 0.877 1.