Testing the detection and discrimination potential of the new Landsat 8 satellite data on the challenging water hyacinth (Eichhornia crassipes) in freshwater ecosystems

Detecting and monitoring the spatial distribution, configuration and propagation rates of aquatic water weeds (i.e. water hyacinth) in freshwater ecosystems, is arguably important to water resources managers, hydrologists and policy makers, for sustainable water resources management and planning purposes. The principal objective of this work was to develop a model that can satisfactorily detect and map the spatial distribution and configuration of water hyacinth (Eichhornia crassipes) in freshwater ecosystems, using both visible, near-infrared and thermal band information derived from the Landsat 8 multispectral sensor. Statistical analysis was done, using two machine learning classification ensembles, which are the Discriminant Analysis (DA) and Partial Least Squares Discriminant Analysis (PLS-DA). Our results have shown that the spatial distribution and configuration of water hyacinth can be accurately detected and mapped with an overall classification accuracy of 95% using Landsat 8 data. The results have further shown that the different growing stages (i.e. young, intermediate and old water hyacinth) could be spectrally detected and mapped, using the new Landsat 8 sensor. In addition, the findings of this study have demonstrated that Landsat 8 bands 5, 6, 7, 8, 10 and 11 are the most influential in detecting and mapping water hyacinth in freshwater ecosystems. Furthermore, allocation of agreement results showed that the DA classification algorithm outperformed PLS-DA in the detection and mapping the distribution and spatial configuration of water hyacinth in freshwater ecosystems. Overall, the derived water hyacinth maps provide critical information required for the development of effective and robust water hyacinth control and eradication programmes.