Investigating the Potential for the Detection of African Lovegrass and Bitou Bush Using SkySat Earth Observation Satellites

In recent years, the use of imagery collected using unmanned aerial vehicle (drones) based imagery and analysed using machine learning algorithms has grown in popularity for invasive plant detection and management. However, drone‐based management of species such as African lovegrass ( Eragrostis curvula ) and bitou bush ( Chrysanthemoides monilifera ssp. rotundata ), has been logistically challenging due to infestations occurring at large scales in mixed landscapes. Earth Observation satellites could provide a more cost‐efficient method for the detection and monitoring of African lovegrass and bitou bush, but this approach currently remains under‐explored. Planet's SkySat satellites produce high‐resolution imagery (0.50 m/pixel), increasing the potential for detecting species in mixed landscapes at scale, compared to moderate‐resolution sensors such as those on Sentinel‐2 and the Landsat systems. This study explored the application of SkySat satellite imagery and support vector machine and random forest classification algorithms for the detection and mapping of African lovegrass and bitou bush. The random forest algorithm consistently produced the highest accuracies, at 89.8% for African lovegrass and 86.1% for bitou bush. The red–blue normalised index, spectral clusters measuring landscape heterogeneity and green spectral band were important predictors for African lovegrass, while blue and red bands and spectral texture through grey‐level co‐occurrence matrix were significant for the detection of bitou bush. Although the challenges associated with mixed species were evident in this study, the emergence of ultra‐high Earth observation satellites, such as SkySat, can be a significant enabler in the cost‐efficient, rapid detection and monitoring of invasive plants in Australia.