Real Time and Near-Real Time Social Research on Operational Forecasting during Severe Weather Events: Lessons Learned from VORTEX SE 2016

This paper describes the use of multiple qualitative, ethnographically-informed social scientific research methods as they were deployed in real time and near-real time to study operational forecasters and forecasting during the Spring 2016 Verification of the Origins of Rotation in Tornadoes Experiment-Southeast (VORTEX-SE) operational period. Drawing on the lead author’s experiences during 8 Intensive Observation Periods (IOPs) to Northern Alabama as part of the 2016 VORTEX-SE team, we will present findings that compare pre-event and post-event data from operational forecasters regarding specific targeted severe weather events with the empirical observations gathered during those events (both social and physical science observations). Special attention will be paid to the challenges that researchers must be aware of regarding the timing of data gathering (real time vs. near-real time) and how the differences that can arise must be accounted for in data analysis. As such, this presentation will contribute both 1) important observations on the challenges forecasters in the southeast face in preparing for, working, and learning from severe weather events and 2) important findings regarding novel methods and methodological challenges that arise in social scientific research on operational forecasters and forecasting.

We observed that forecasters in the US face unique challenges in forecasting tornadoes and tornadic events related to the complexity of severe weather events in the U.S. southeast. This complex environment included: rapid storm evolution; complex interaction effects with conditions in the Gulf of Mexico; increased risks associated with difficult to predict, high shear, low CAPE events; poorly understood interactions between storms and mesoscale events and terrain in the southeast; etc. In addition to these forecasting challenges, the southeast U.S. also is different from other tornado-prone regions in the country in its demographic composition, which contributes to added communication and warning challenges – population density is higher than much of the Great Plains; the southeast population is often less well-informed regarding protective actions and preparedness; there is a higher proportion of at-risk housing stock; terrain complicates the capacity to inform the public because it obscures visual confirmation of storms; etc.

Social scientists have studied the nature of operational weather forecasting as a discrete instance of “science in practice.” As part of the 2016 VORTEX-SE team that conducted research in Northern Alabama during the Spring 2016 severe weather season (March-May 2016), we describe how near-real time and real time interviews and in situ operational observation combined to provide a unique understanding of operational forecasters facing the unpredictable severe weather challenges experienced in the southeast U.S. Real time in situ observation of operations allowed researchers to record interactions between forecasters, the flow of information in the operations areas, and how decisions were made and priorities set within the WFO during severe weather events. Near-real time data gathering involved conducting pre- and post-event interviews with operations meteorologists (pre-event interviews were conducted between 8-36 hours before an event; post-event interviews were conducted within 24 hours of the end of an event). These pre- and post-interviews focused on four key topics: 1) what was/would be most difficult to predict about an event, 2) what they were most confident about regarding an event, 3) how they judged model guidance for an event pre- and post-event, and 4) overall impressions, descriptions, and predictions related to an event. In addition, deep background interviews were conducted approximately one month prior to the beginning of the 2016 VORTEX-SE field experiments season, permitting the researchers to gather information on the professional background, views about various aspects of forecasting (the use of predictive numerical model guidance, the reliability of observational data, etc.), and beliefs regarding communicating risks and weather information to local “publics.” In addition, background interviews were used to gauge the forecasters’ general perceptions of the “uniqueness” of severe weather in the southeast U.S. In total, 100 interviews were conducted (14 background (average 75 minutes), 4 ad hoc interviews (average 20 minutes), 82 pre-/post-event interviews (average 18 minutes)) in addition to the recording of several hundred severe weather event-related social media posts, and observation notes on operational activities recorded during each of 8 IOPs (1 pre-VORTEX, 7 during VORTEX).