An Extensible Tool for Estimating Space Weather Benchmarks

Space weather benchmarks are recognized as a crucial product for a variety of government and industry stakeholders. Until now, they have been computed primarily from a scientific perspective, on an individual basis, and, to a large extent, without cross-validation. We propose to develop a tool that combines the most robust techniques currently available, together with a wide range of data from the Heliophysics System Observatory (HSO), and the option to upload user-supplied data, to produce the most accurate estimates of benchmarks, together with their uncertainties. During our Phase I effort, we developed a prototype web-based tool that illustrates how space weather benchmarks can be estimated using the most sophisticated statistical methodologies, which have been previously investigated by our team. Importantly, we carefully quantified the uncertainties with these benchmark estimates, a quantity that is at least as useful as the actual benchmark value. We incorporated a selection of sophisticated methodologies (e.g., Peaks-Over-Threshold). In Phase II we propose to refine this tool substantially by: (1) carefully migrating all code from R to Python; (2) migrating the Shiny app to Dash; (3) incorporating a range of new numerical techniques into the pipeline; (4) Generalizing the analysis to include spatially resolved datasets; (5) Adding multivariate analysis; and (6) Providing additional refinements to the model fitting. We anticipate that this tool will find broad appeal within the NASA community, and, ultimately, across many other scientific and engineering disciplines where an accurate assessment of risk likelihood is necessary. We plan to commercialize this work by providing tailor-made solutions for customers, including support and service.nbsp;