Accurately mapping understory vegetation using optical remote sensing images has been challenging because of canopy obstruction. To address this challenge, a novel method was proposed to identify the staple-food bamboo of the giant panda (hereafter SFB-GP), a typical understory species. Leveraging space-air-ground integrated monitoring (SAGIM) technology, the SFB-GP was accurately identified in large areas. First, precise plot-scale SFB-GP reference maps were produced using multitemporal UAV-RGB observations, which were aggregated to create training samples. A regional-scale mapping model was then developed using time-series Sentinel-2 imagery and a random forest algorithm, incorporating an optimized selection of spectral features, vegetation indices, phenological features, and terrain features. Taking the Wanglang National Nature Reserve (WNNR), a core zone of Giant Panda National Park (GPNP), as the study area, the spatial distribution of the SFB-GP across the WNNR was extracted, achieving an overall accuracy of 83%. Phenological differences between the SFB-GP and canopy vegetation play a critical role in SFB-GP mapping, with key phenological periods occurring during the non-growing season. This study provides not only high-resolution SFB-GP distribution data for the conservation of the giant panda but also a methodological framework for SFB-GP mapping across the entire GPNP and other types of understory vegetation.