On the Use of UAS for Prediction of Severe Convective Storm Initiation

The “convection initiation” (CI) problem consists of the need to observe pre-storm moisture and temperature fields and the storm-relative flow with spatial and temporal resolutions of approximately 30 km and 30 min. It is also desirable to obtain measurements of the strength and depth of mesoscale updrafts, in order to determine whether air parcels reside long enough within the updrafts to reach their “level of free convection” before leaving the updrafts. While weather radar has clearly been proven invaluable for the detection of severe storms, and is also the primary observing system utilized in numerical weather prediction models dedicated to their short-term prediction, radar cannot measure these critical CI fields. Ground-based remote sensing systems have been suggested as a viable solution to this need, but their cost makes deployment of large networks of such instruments currently impractical. Geostationary satellite infrared sensors lack the needed vertical resolution in the boundary layer to determine this information with sufficient fidelity and are limited by cloud cover.

Low-altitude, short-endurance Unmanned Aircraft Systems (UAS) can help fill this data void. The concept we have been exploring is that of profiling, rotary-wing UAS to obtain vertical profiles of winds, temperature, and humidity within the lowest 2,500 ft of the atmosphere every 30 min at fixed locations, in combination with fixed-wing observing systems between the copter sites in order to measure the spatial heterogeneity of the atmosphere between the sites. In the presentation, results will be shown validating UAS sensor measurements against data from mobile sounding systems and ground-based remote sensing systems (Doppler wind lidar, sonic anemometer, and infrared interferometer profiling systems). We will also discuss the value of these data as determined by National Weather Service (NWS) forecaster evaluation in a real-time experiment held in May 2017.

The ultimate goal is to provide targeted, adaptive, and fully autonomous sampling within FAA-approved regions that would be executed by the NWS as needed when severe weather threatens. The desired outcome would be that the data could be shown useful for initializing high-resolution “Warn On Forecast” models aimed at making a one hour tornado warning a reality. We will discuss the current regulatory limitations preventing realization of this ultimate objective, but also what has been learned with current systems and steps to be taken toward this goal.