Development of a Predictive Tool for Tornadoes Downwind of the Appalachian Mountains Using AO and NAO Indices

The majority of current tornado research focuses on the Midwest, leaving a gap in our understanding of how and why tornadoes form in the eastern United States. In order to help bridge this gap, the influence of the Artic Oscillation (AO) and North Atlantic Oscillation (NAO) on the prevalence of tornadoes downstream of the Appalachian Mountains was analyzed in an effort to develop an innovative forecasting tool. This was performed using tornado reports on the lee side of the Appalachians from 1950-2015, for the months of March through August. These reports were consolidated into unique events, each assigned a count indicating the number of tornadoes associated with that particular event. Daily AO and NAO index values were acquired for the two weeks preceding each event, and time series analyses were performed on these precedent indices. Results show that, on average, AO and NAO indices reach a maximum roughly four and seven days, respectively, prior to downstream tornado events, before decreasing rapidly up through the events themselves. Furthermore, the magnitude of both these maxima and the subsequent drop-off increases with the number of tornadoes associated with the events, and is maximized for the categories of events with 10-15 tornadoes. These findings suggest that the synoptic setup associated with the maximized AO and NAO teleconnections creates an environment downstream of the Appalachian Mountains that is particularly conducive to the development of tornadic supercells, potentially through regional tropospheric destabilization via moisture and warm air advection. The sudden decrease in the AO and NAO indices corresponds to a significant shift in the given synoptic pattern, which combined with local mesoscale features, provides sufficient forcing for the initiation of lee side convection. AO and NAO indices can be accurately predicted out to at least a week, so the identification of an index pattern similar to one described above could signal an increased likelihood of tornadic development downstream of the Appalachian Mountains, and therefore be used as a diagnostic tool for operational forecasters. It was found, however, that this pattern alone is not sufficient to predict tornadic events. Thus, the next step in the research will be to incorporate data from the National Climatic Data Center (NCDC) to determine if another meteorological variable can help distinguish between those synoptic environments that are favorable for tornadic development and those that are not. Of particular interest is dewpoint depression. This quantity can be used a proxy for LCL height, which has been shown in previous literature to have an impact on the low level characteristics necessary for tornado formation. Once NCDC data has been integrated into this analysis, several case studies will be performed to demonstrate the applicability of the forecasting methods developed.