Spectral Algorithm Framework

The Remote Detection Program within the Office of Defense Nuclear Nonproliferation Research and Development is tasked with developing new technical capabilities to improve the detection, identification, and quantification of special nuclear and radiological materials in order to detect new nuclear threats and assure nuclear test treaty compliance. Due to limited access to acknowledged foreign sites, remote detection of illicit behavior is becoming increasingly critical for mission success. Hyperspectral Imaging is a modality that generally enables remote characterization of a material by measuring its radiance return as a function of wavelength, which is distinct for every chemical. Algorithms to separate the target source from the background and atmosphere allow users to deduce the chemical makeup of ground materials and their quantities in order to detect manufacturing processing and infer quantities of licit or illicit materials being produced. The Phase I and initial Phase II of this project focused on developing a variety algorithms for detection, identification, and characterization of materials and a integrating them into a commercialized workflow optimization framework. One class of algorithms extended hyperspectral imaging capabilities from visible wavelengths through long-wave infrared for both solid and thin-plume gaseous targets. Novel atmospheric compensation methods were studied and implemented, focusing on the purely reflective and the purely thermal regimes. Critically, extensive studies of the effects of errors in atmospheric compensation and background subtraction on quantification and detection metrics were completed in all wavelength regimes. Plans for the Phase IIA project consist of extending algorithms completed in Phase II into other mission-relevant regimes, including detection of optically thick gaseous plumes and atmospheric compensation in the mid-wave infrared in which there is a mix of reflective and thermal radiance. In addition, co-indication analysis originally proposed for Phase II will be studied using a mix of liquid and gaseous effluents using extensive Deepwater Horizon Oil spill datasets. This will require the implementation of thin-film liquid detection algorithms. The benefits of this improved framework with measurements in the spectrally rich mid-wave infrared region as well as co-indication analysis of all types of effluents will be realized in a further strengthening of capabilities in confirming the presence of nuclear threats. The phenomenology developed for the detection of thick gaseous plumes will also allow for the determination of the extent of nuclear fallout particles in the event of a nuclear/chemical weapon attacks. In addition to non-proliferation efforts, commercialization of the framework in Phase III and beyond can apply directly to programs across the intelligence community from agricultural, environmental and climate monitoring, detection of illegal mining activities, or the discovery of illicit activities such as drug manufacturing by local law enforcement.