Assessment of airborne remote sensing data for high-resolution mapping of invasive Prosopis spp. in a semi-arid environment in Kenya

Invasive alien plant species are a major global problem threatening biodiversity and livelihoods and their mapping is needed for understanding their distribution dynamics, and for facilitating control and eradication measures. Prosopis spp., a fast-growing woody species native to South America, have been widely introduced into the tropics to restore degraded areas, but they have spread uncontrollably. For example, in East Africa, Prosopis spp. have invaded rangelands and thus decreased plant diversity and affected the livelihoods of pastoral communities. Remote sensing instruments mounted on an aircraft can be used to map such species and especially a combination of different sensors holds a potential for accurate detection.

The objective of this study was to test how a combination of airborne light detection and ranging (LiDAR), hyperspectral, and fine resolution multispectral data can be used to map Prosopis spp. in a semi-arid environment in Kenya. The remotely sensed spectral, structural, and textural features were used in a one-class machine learning algorithms to detect these species in a complex landcover. The results provide information on the use of different airborne remote sensing instruments and their combination in mapping woody alien invasive species and offer insights on the distribution of Prosopis spp. in the study area.