Empirical validation of a hyperspectral systems model for subpixel target detection using data from a new UAS field collection

Systems modeling of hyperspectral instruments is effective for forecasting instrument performance of subpixel target detection. A new reconfiguration of the statistical model-based tool FASSP (Forecasting and Analysis of Spectroradiometric System Performance) is currently in development at the Rochester Institute of Technology, for purposes of exploring systems limitations in subpixel detection. To validate the baseline functioning of the statistical model, empirical analyses using real data were cross-examined with model predictions. The real data were collected from a field experiment using a hyperspectral sensor on-board an unmanned aerial system (UAS). To assist in model validation, a variety of novel subpixel targets, spanning a range of constant target percentages, were designed and deployed in the field. The UAS was advantageous in enabling the research team to maintain full end-to-end control of the system parameters within the experiment. This includes selecting specific flight lines, collecting ground truth spectral measurements, deploying specific targets, and processing raw data into geophysical units of surface reflectance. The study revealed close alignment between the empirical results and modeled predictions.