Using remote sensing to estimate the impacts of wattle species on native grass vegetation.

Abstract

This study was stimulated by the long standing challenge of the lack of suitable satellite data with optimal temporal, spectral, and spatial resolutions to monitor rangelands. The study, therefore, sought to evaluate the utility of remotely sensed data in estimating the impact of wattle infestation and clearance on native grass species productivity and diversity. The first objective of this study was to investigate the utility of Sentinel 2 Multispectral Imager (MSI) remotely sensed data and Partial Least Squares regression as a cost-effective and quick assessment technique to map above ground biomass (AGB) of native grass growing under different levels of Acacia baileyana, A. dealbata & A. mearnsii, invasion in Matatiele, South Africa. The second objective focused on assessing the impact of wattle invasion on grass species diversity. This was achieved by investigating the utility of Sentinel-2 MSI data in optimally estimating grass Species richness, Shannon Wiener and Simpson’s diversity indices at different levels of wattle invasion. In relation to the first objective, the findings of this study showed that Sentinel 2 MSI data derived vegetation indices optimally estimated biomass in relation to standard wavebands. Results also showed that Sentinel 2 MSI data (combination of raw spectral bands and vegetation indices) predicts grass AGB levels of wattle invasion at reasonable accuracies (RMSE = 19.117g/m2 and R2 = 0.8268). The most influential variables in estimating biomass across different levels of wattle invasion were red edge based vegetation indices (VIs) and bands 5,6 and 7. With regards to the second objective, this study showed that following restoration, there were no significant difference (p > 0.05) between cleared and uninvaded grassland areas. Results also showed that diversity indices were optimally modelled when compared to species richness. However, for all three diversity variables, individual raw spectral bands yielded lower accuracies when compared to vegetation indices. Overall, the most influential spectral variables were, bands 5 and 6, NDVI computed from bands 6 and Band 3. Results of this study also showed that Shannon Wiener’s index better predicted grass species diversity across different levels of wattle invasion in an alpine grassland (RMSE = 0.2145, R2 = 0.6392) in relation to the other diversity indices. This study was able to demonstrate that Sentinel-2 MSI spectral variables have a potential of offering reliable and accurate estimates of grass species diversity in a wattle infested grassland. The study therefore advocates for the utility of remotely sensed data in monitoring grassland degradation and restoration.