Enabling probabilistic retrospective transport modeling for accurate source detection

Abstract

Predicting source or background radionuclide emissions is limited by the effort needed to run gas/aerosol atmospheric transport models (ATMs). A high-performance surrogate model is developed for the HYSPLIT4 (NOAA) ATM to accelerate transport simulation through model reduction, code optimization, and improved scaling on high performance computing systems. The surrogate model parameters are a grid of short-duration transport simulations stored offline. The surrogate model then predicts the path of a plume of radionuclide particles emitted from a source, or the field of sources which may have contributed to a detected signal, more efficiently than direct simulation by HYSPLIT4. Termed the Atmospheric Transport Model Surrogate (ATaMS), this suite of capabilities forms a basis to accelerate workflows for probabilistic source prediction and estimation of the radionuclide atmospheric background.