Satellite-imagery-based landscape monitoring is a widely used technique for biodiversity analysis. The free access to multi-source satellite data with different technical characteristics has facilitated ecological landscape characterization. However, there are no unified approaches for harnessing the combined potential of these data sources. Despite the emergence of satellite imagery fusion techniques, challenges persist regarding relevant information loss. In response to this challenge, we propose a novel methodology for not combining raw images but their features, such as spectral ecological indices, using traceable linear data fusion techniques. As a proof-of-principle, we test it by combining features of four satellites (one active) to delineate the boundaries of tropical forested areas automatically. We achieved up to 90% accuracy in classifying grass, tall grass, and forested zones. This approach enhances the characterization of ecological landscapes while minimizing information loss and improving the ecological interpretability of the results.