The accelerating spread of invasive Acacia species poses a severe threat to native ecology, ecosystem services and environmental sustainability in the Cape Floristic Region (CFR), which is recognised as one of the biodiversity hotspots of the world. As such, effective management of these threats relies on accurate and timely monitoring of invasive plant species. Remote sensing has been widely used to map invasive species; however, the emergence of new spaceborne hyperspectral sensors necessitates the evaluation of their capabilities. This study investigated the potential of the Environmental Mapping and Analysis Program (EnMAP) hyperspectral and Sentinel-1 radar data to discriminate between two invasive Acacia species (A. cyclops and A. mearnsii). Four data scenarios were evaluated: (1) spectral bands alone, (2) spectral bands combined with radar data, (3) EnMAP-derived principal components (PCs), and (4) PCs combined with radar data. The eXtreme Gradient Boosting (XGBoost) algorithm was used to classify the two Acacia species using each of the four data scenarios as predictors. Integrating radar data with spectral bands improved the classification accuracy from 83% to 85% for A. cyclops and from 87% to 88% for A. mearnsii, compared to using spectral data alone. PCs of EnMAP bands slightly reduced the accuracy when compared to using spectral data alone, with the producer’s accuracy for A. cyclops decreasing by 2% (from 83% to 81%) and for A. mearnsii by 5% (from 87% to 82%). However, the addition of radar data to the PCs improved the accuracy, increasing the producer’s accuracy for A. cyclops by 2% (from 81% to 83%) and for A. mearnsii by 2% (from 82% to 84%). The visible and shortwave-infrared (SWIR) bands of EnMAP had high contributions to the identification of the species. In general, the study showed the capability of EnMAP hyperspectral data and radar data for mapping Acacia species.
Matsokane et al. (Wed,) studied this question.