Discriminating Rangeland Management Practices Using Simulated HyspIRI, Landsat 8 OLI, Sentinel 2 MSI, and VENµS Spectral Data

This study sought to spectrally discriminate grasses grown under different management practices (i.e., mowing, grazing, fertilizer application, and burning), using field spectrometer data resampled to Hyperspectral Infrared Imager (HyspIRI), Landsat 8 Operational Land Imager (OLI), Sentinel 2 Multispectral Instrument (MSI), and Vegetation and Environment monitoring on a new MicroSatellite (Venμs). The study is inspired by the long standing challenge of lack of suitable satellite data with high temporal, spectral, and spatial resolutions for rangelands monitoring. Specifically, this study spectrally discriminated grasses grown under 1) different rangeland management practices as well as 2) different levels of application of each practice. Results of this study show that the spectral setup of HyspIRI, Sentinel 2 MSI, Venus, and Landsat 8 OLI yielded high accuracies of up to 92%, 82%, 83%, and 75% overall accuracy, respectively, in discriminating grass grown under different rangeland management practices. The high classification accuracies were exhibited by the use of vegetation indices and wavebands located in the red edge (HyspIRI: 700, 740, and 780 nm and Sentinel 2 MSI: bands 5, 6, and 7 8a) and NIR (HyspIRI: 700, 740, and 780 nm and Sentinel 2 MSI: band 8a) spectra, respectively. Results of this study illustrate that although simulated Sentinel 2 MSI data yields lower classification accuracies when compared to HyspIRI, it offers better classification accuracies with high agreements between training and testing datasets when compared to the HyspIRI data. Overall, the findings of this study underscore the potential of upcoming satellite missions in ensuring informed rangeland monitoring and management applications.