The Development of A Regional, HIGH-RESOLUTION OSSE Framework To Study The Impact of Observations From Uncrewed Aircraft Systems On Numerical Weather Prediction

Abstract

Using uncrewed aircraft systems (UASs) for routine weather observations is becoming increasingly feasible, but two open questions are the impact UAS observations will have on operational NWP and the optimal configuration of a UAS observing network. Deploying hundreds to thousands of UAS across the United States to answer these questions is not feasible, so we propose using an observing system simulation experiment (OSSE) instead. This article describes the development and validation of an OSSE framework for examining the impact of UAS observations on regional, convection-allowing NWP. The OSSE includes two weeklong nature runs (NRs) over CONUS with 1-km grid spacing, simulated conventional observations, a 3-km forecast system similar to the prototype Rapid Refresh Forecast System (RRFS), and verification workflows. Comparisons between the NR and various observations show that the NR generally mirrors reality, though the NR tends to produce too many reflectivity objects that are too large and a larger coverage of intense precipitation rates compared to reality. Comparing real-data and OSSE RRFS runs indicates that, for the majority of the variables examined, forecast errors are not systematically smaller in the OSSE, suggesting that there is no identical twin issue. Data-denial experiments using either real or simulated conventional observations indicate that the OSSE has a similar ordering of the most to least impactful observations, though the impact of aircraft observations tends to be larger in the OSSE. Altogether, these results suggest that the OSSE framework can be used to glean meaningful information about UAS observation impacts on regional NWP.