Hyperspectral imaging from a light-weight (up to 75 kg) unmanned aerial vehicle platform

Since 2009 the Idaho National Lab (INL) has been developing advanced remote sensing capabilities that combine increasingly sophisticated miniaturized sensors with relatively affordable, light weight (under 75 kg) unmanned aerial vehicles (UAVs). UAV-based hyperspectral sensing capabilities have been routinely refined via flight tests conducted at INL's UAV Runway Research Park in southeastern Idaho, and at the Orchard Training Area in central Idaho. Idaho State University (ISU) Boise Center Aerospace Lab (BCAL) has provided field data collection and image processing support to target ground versus aerial data comparisons, assess spectral and geometric data accuracy and determine classification algorithms appropriate for vegetation management applications. We report instrumentation, sensor and image validation results, optimal flight parameters, and methods for improving the geometric accuracies of the datasets. We also assess the accuracy of narrowband vegetation indices and shrub cover estimates derived from the imagery. Preliminary results indicate that the UAV-based hyperspectral imaging system has potential to bridge the gap between costly in-situ data collections, coarse resolution satellite data collections, or infrequent and costly manned hyperspectral data collections. Furthermore, new areas of research may be possible with this UAV platform by providing an affordable, on-demand platform that can rapidly collect transect data and stay on station for hours.