Indonesia's forests are undergoing rapid changes due to land cover transformation, presenting challenges for monitoring these ecosystems, particularly in areas with mixed land cover and ongoing deforestation. This study aims to map 30 m fractional tree cover (FTC) in Indonesia across four forest types: secondary dryland forest, secondary swamp forest, secondary mangrove forest, and plantation forest. It leverages spectral-texture data derived from Landsat-8 images, including spectral bands, vegetation indices (VIs), tasseled cap transformation (TCT), gray-level co-occurrence matrix (GLCM), and geometric features, and integrates these with PlanetScope-3B (PS-3B) images for their superior spatial resolution. Tree-based pipeline optimization tool (TPOT) models were employed to establish relationships among these features for estimating FTC. The models demonstrated high accuracy on validation data, achieving coefficients of determination (R2) values of 0.96, 0.98, 0.96, and 0.95; root mean square error (RMSE) values of 0.09, 0.08, 0.13, and 0.12; and mean absolute error (MAE) values of 0.05, 0.06, 0.10, and 0.09 for the four forest types, respectively. When validated with aerial images, the models achieved R2 values of 0.85, 0.85, 0.85, and 0.93; RMSE values of 0.16, 0.12, 0.10, and 0.14; and MAE values of 0.13, 0.09, 0.08, and 0.12. The model applied to the secondary mangrove forest was also validated with 44 independent ground measurement data, achieving an R2 of 0.80, an RMSE of 0.07, and an MAE of 0.06. A comparative analysis with global FTC products revealed the highest consistency with the Global Forest Watch (GFW) product, with an R2 of 0.72, an RMSE of 0.15, and an MAE of 0.12. Field checks confirmed that the results closely align with actual conditions, underscoring the robustness of the proposed approach. This study concludes that integrating PS-3B images with Landsat-8 data, combined with TPOT models, offers an innovative way to map FTC across global ecosystems.