Exploring impacts of radar scan time on NWS warning decision making

The S-band phased-array radar (PAR) at the National Weather Radar Testbed (NWRT) in Norman, Oklahoma electronically scans storms with high-temporal resolution volumetric sampling (< 1 min). While some NWS forecast offices have had access to frequent low-altitude updates (~1 min) with Terminal Doppler Weather Radar, none have had routine access to rapid-update volumetric data. Though previous Phased-Array Radar Innovative Sensing Experiments (PARISE) have introduced NWS forecasters to the rapid-scan capability of PAR, the impacts of such data on warning decision making and warning lead time have not been examined.

The key objective of the 2010 PARISE was to develop and pilot the first comparative study designed to examine and quantify potential impacts of rapid-update phased-array radar (PAR) data on NWS warning decisions and warning lead time. The 12 participating forecasters included 3 females and 9 males from NWS offices located in 11 different states, most east of the Rocky Mountains. In each of three experiment weeks, four forecasters acquainted themselves with NWRT PAR data and WDSS-II display software through several types of events on Tuesday and Wednesday. On Thursdays, forecasters worked in pairs that had approximately equivalent radar interrogation skills through two quickly evolving, marginal tornado events. One pair had full-temporal NWRT PAR data, while the other had NWRT PAR data with data removed to approximate the time resolution of WSR-88D. Each team issued any warnings, warning updates, and cancellations they deemed appropriate. After each event, each team reflected upon their actions with a facilitator before coming together to discuss the dependence of their warning decision making on temporal resolution. Additional data include observation notes, video of weather radar and other observations viewed and analyzed, and recorded discussion between forecasters as they worked. The video of forecasters analyzing weather radar data has been crucial for determining their warning decision process.

The 12 participating forecasters agreed when using the rapid-update data it was far easier to apply conceptual models and there was much less uncertainty regarding storm severity and evolution. Because these data were new, participants reported they continued to adapt their normal approach to interrogating radar data throughout the week. Some discovered the rapid updates allowed easier use of vertical cross sections. We continue to analyze the impact on warning decision making, seeing a variety of impacts. In one sample case, the use of rapid-update PAR data resulted in significant improvement in tornado warning lead time: 21 min for the forecaster pair with 43-s updates, and 0 min for the forecaster pair with 4.5-min updates. The 0-min lead time was due, in part, to software issues. In comparison, the forecaster pair with 43-s updates was more decisive about which storm to issue a warning on, observed more persistent features aligned with their conceptual model than the other forecaster pair, and had more confidence in their understanding of the event.