OSSE simulations of targeted observations of supercell thunderstorms by unmanned aircraft systems

Supercell thunderstorms are believed to produce a disproportionate share of severe weather and are likely the parent storms for nearly all significant tornadoes. Although the precipitation field associated with supercells is typically well-resolved in surveillance radar data, accurate real-time predictions of supercells are ultimately confounded by inadequate initial conditions. Such real-time predictions, like those envisioned as part of a warn-on-forecast system, have the potential to revolutionize the National Weather Service warning decision process, so it is vitally important to consider ways to modernize the weather surveillance networks to fill this data hole.

UAS have the potential to play an important role in future weather surveillance networks. In particular, the low-level thermodynamic characteristics of supercell thunderstorms are rarely sampled, despite mounting evidence that they are crucial in determining the tornado potential of storms. Through targeted surveillance, low-level in situ observations of the storm and proximity environment collected by UAS could be assimilated into storm-scale simulations for use in the warning decision process. The primary objective of this work is to determine which regions of a supercell should be sampled by unmanned aircraft in order to provide the greatest improvements to storm-scale model forecasts. The observing system simulation experiment approach is utilized to conduct this assessment.

In this study, low-level observations are collected by a simulated aircraft within a nature run (100 m horizontal grid spacing) of an idealized supercell. The observations are assimilated using the Data Assimilation Research Testbed into a degraded, coarse run (1 km horizontal grid spacing) of the same idealized supercell, to approximate models that will be used in early warn-on-forecast systems. Prior work on targeting observations has used the ensemble sensitivity theory to predict which regions of meteorological phenomena to sample to obtain the greatest improvement in the forecast. Ensemble sensitivity theory predicts a linear relationship between the expected forecast response by perturbing the model simulation and the actual response from assimilating observations. Although this relationship appears to be valid for synoptic scale phenomena, recent work has demonstrated that ensemble sensitivity theory may not be appropriate for mesoscale phenomena. In addition to identifying which regions to sample, another goal of this study is to test the validity of ensemble sensitivity theory in forecasting supercell thunderstorms. Analysis of preliminary results is underway and final results of the simulations and their implications on UAS targeted surveillance of supercells will be presented at the conference.