Investigating the Impacts of Assimilating Surface Observations on High-Resolution Forecasts of the 15 May 2013 Tornado Event

In this study, the Advanced Regional Prediction System (ARPS), and its associated three-dimensional variational analysis (3DVAR) package are used to simulate a tornadic supercell at 400-m grid spacing. This storm produced an EF3 tornado in Johnson County, TX during the evening of 15 March 2013. Data from Doppler radar, satellite, aircraft, radiosondes, profilers, and surface observations are assimilated in this work. We show that the assimilation of non-conventional surface observations from three networks: the Citizen Weather Observer Program (CWOP), Global Science and Technology (GST) and Automated Weather Stations (AWS) operated by EarthNetworks, in and around the storm inflow, were fundamentally important to the development of an intense low-level mesocyclone. Simulations that did not incorporate this non-conventional data either developed a weak mesocyclone that was displaced to the east of the actual tornado track, or were unable to develop a defined mesocyclone in the first place. In particular, the assimilation of thermodynamic variables (temperature, pressure and moisture) from a subset of AWS observations near the storm’s inflow environment leads to the most accurate simulation of the low-level mesocyclone.