Fires in tropical regions such as the Amazon, Congo, and Indonesia are human-caused and threaten native ecosystems - especially when areas burn multiple times within several years and during drought conditions. Satellite data is typically the first line of information about these fires, but they can be particularly hard to detect and track from space. Current datasets underestimate active fires and burned areas due to factors like cloud cover or tree canopies blocking the signal, or fires being too small to detect. Our work seeks to improve tropical fire detection and tracking by leveraging multiple satellite-based sensors and increasing algorithm sensitivity to fire energy from small or obscured fires in different land cover types.