Observing system simulation experiment studies on the use of small UAV for boundary-layer sampling

Abstract

A long-desired component to the U.S. operational observing systems is the capability to measure vertical profiles of wind, temperature and moisture in the lower troposphere at high spatial and temporal resolution. This study proposes that such profiling could be done by small unmanned aerial vehicles (UAVs) assuming that autonomous flights at least through the depth of the boundary layer be permitted. Since we do not yet have FAA permission to test such an observing network, we examine the potential improvement that a UAV network could have on storm-scale numerical weather prediction using an Observation System Simulation Experiment (OSSE) approach. An OSSE is performed over the state of Oklahoma in which we assume that a UAV could be launched from 110 Oklahoma Mesonet stations every hour, fly vertically to an assigned maximum altitude and return to its charging station, providing soundings at a roughly 35 km horizontal resolution. We begin with a case study of convective initiation (CI) as a compromise between a fair weather day and one with extensive ongoing convection. The OU ARPS model provides a nature run at high (1 km) resolution, while the control run and OSSE experiments are done with the WRF-ARW model at 3 km. To simulate the effect of data from dozens of observing systems already included in operational models, the nature run data volume is sampled at synoptic scales and inserted into the control run via a 6-hr data assimilation (DA) period. Simulated hourly UAV temperature, moisture and wind data, with expected errors, are then added to the DA, followed by 12-hr forecasts. The analyses and forecasts are examined to assess the added value of UAV data. Tests are run to measure the impact of varying the maximum UAV altitude and the spatial density of UAV observations. Initial results clearly show an improved boundary layer structure and subsequent CI location and timing when UAV data are added to the control experiment. Additionally, early findings indicate flight altitude and network density can play a role in the quality of the DA analysis and subsequent forecast of convective initiation. Although sensitivities to the quality of the moisture analysis are noted, the results here suggest that a real-world deployment of automated UAVs could have a positive impact on atmospheric analyses and short-term numerical weather prediction.