Using an OSSE to Explore the Impact of Assimilating Different Amounts of Uncrewed Aircraft System Observations on Regional NWP Forecasts

Over the last decade, uncrewed aircraft systems (UAS) have seen widespread use in the meteorological research community, with researchers using UAS to collect observations of phenomena ranging from severe local storms to tropical cyclones. With the increasing maturity of UAS for weather applications, there has been interest in collecting routine weather observations from UAS as a way to better sample the planetary boundary layer. Before a nationwide network of UAS can be deployed, however, it should first be determined whether UAS observations will have a positive impact on NWP forecasts and how to best configure a nationwide UAS network. To fill these knowledge gaps, an observing system simulation experiment (OSSE) was designed to assess the impact of UAS on regional NWP forecasts. The OSSE consists of two nature runs, both of which are 1-km, week-long WRF simulations over CONUS. To increase the diversity of meteorological phenomena, each of the two nature runs captures a different season (winter and spring). Simulated commercial aircraft, surface station, radiosonde, GPS-derived precipitable water, and UAS observations are then created from the nature runs and assimilated into a variant of the Rapid Refresh Forecast System (RRFS).

The initial set of experiments focuses on different spatial densities of vertically profiling UAS that fly up to 2 km AGL every hour, with spatial densities varying from 150 to 35 km between adjacent sites. Assimilating UAS observations has a positive impact on RRFS analyses and forecasts, particularly below 600 hPa, with root-mean-square errors being reduced by over 20% for some vertical levels and variables. While the best forecasts are obtained when assimilating observations from the densest network of UAS, the resulting improvements to RRFS forecasts tend to saturate as more UAS are added. To help assess whether these results are model agnostic, experiments will also be presented using a different forecast model that is similar to the High-Resolution Rapid Refresh (HRRR). Overall, these two sets of experiments represent a first step toward determining the optimal configuration for a nationwide network of weather-observing UAS in the United States.