Using Tracer Experiment Data to Test and Evaluate the HYSPLIT Lagrangian Model Inverse System

HYSPLIT is one of the most widely used Lagrangian dispersion models in the atmospheric sciences community. This very versatile modeling tool provides simple trajectories as well as very complex atmospheric processes calculations, using 3-D particles or puffs, that include emissions, transport, mixing, chemical transformations, and deposition of hazardous materials. Recently, an inverse system based on 4D-Var data assimilation and a transfer coefficient matrix (TCM) based on multiple independent HYSPLIT dispersion runs has been developed and successfully used in several applications. In this study, we use the Cross Appalachian Tracer Experiment (CAPTEX) as a dataset for applying the inversion system. These tracer measurements have been extensively used to evaluate various transport and dispersion models and they provide a characterization of the source’s strength and location that offers a unique opportunity to evaluate the estimated emission sources through a top-down approach. Using this inverse modeling setup we first estimate the source strength assuming the source location and temporal release pattern are known. Then, we recover the temporal variations of the releases, assuming the temporal release patterns are unknown. Furthermore, assuming the source location is unknown; we identify the release location. Finally, we use the HYSPLIT ensemble runs to provide uncertainty information regarding the source estimation results.