Bamboo Classification Using WorldView-2 Imagery of Giant Panda Habitat in a Large Shaded Area in Wolong, Sichuan Province, China

A study demonstrates that WorldView-2 (WV-2) imagery can effectively identify small patches of understory bamboo in giant panda habitats, even in shaded mountainous regions with limited training data. The research achieved high accuracy—93.10% user's accuracy and 82.65% producer’s accuracy for bamboo classification—by expanding sample size using principal component analysis (PCA) and applying a geostatistically-weighted k-nearest neighbor (k-NN) classifier, which accounts for spatial correlation between classes. The findings highlight the potential of WV-2 imagery for habitat assessments in the Sichuan Giant Panda Sanctuaries, including Wolong, by overcoming challenges posed by shadow coverage and sparse data.

Tang, Y., Jing, L., Li, H., Liu, Q., Yan, Q., & Li, X. (2016). Bamboo Classification Using WorldView-2 Imagery of Giant Panda Habitat in a Large Shaded Area in Wolong, Sichuan Province, China. Sensors, 16(11), 1957. https://doi.org/10.3390/s16111957