Exploring the Value of Spire GNSS Radio Occultation Bending Angle Assimilation for Improving HWRF Model Forecasts of Atlantic Hurricane Intensity Weather and Forecasting

Global Navigation Satellite System (GNSS) radio occultation (RO) profiles provide atmospheric temperature and water vapor information, complementing microwave and infrared satellite radiances by enhancing coverage in data-sparse oceanic regions due to their high vertical resolution and all-sky observability. This study uses data denial experiments to evaluate the impacts of assimilating commercial RO profiles from Spire Global, Inc. (Spire) on Hurricane Weather Research and Forecasting (HWRF) model analyses and forecasts of four 2022 Atlantic hurricanes. HWRF currently assimilates RO data from government-funded missions such as the Constellation Observing System for Meteorology, Ionosphere, and Climate-2 (COSMIC-2). Compared to COSMIC-2 observations, Spire data exhibit similar retention statistics under HWRF’s quality control screens and similar diagnosed observation uncertainties, except around the extratropical boundary layer where Spire’s data retention is improved, and its observation errors are smaller. Compared to a Control experiment that withholds COSMIC-2 and Spire data, a C2 experiment assimilating COSMIC-2 observations yields a ~ 0–10% reduction in mean absolute minimum central sea-level pressure (PMIN) intensity errors for most forecast times. A C2Spire experiment assimilating Spire RO data on top of COSMIC-2 generates mean PMIN absolute errors similar to C2 except for the t = 24–36 h period when they are improved over Control by a statistically significant 10–15%. Control’s 3–6 hPa mean PMIN over-intensification bias is reduced by a statistically significant 17–38% in t = 66–108 h C2Spire forecasts. Additionally, short-range C2Spire forecast temperature and specific humidity root-mean-squared deviations (RMSDs) against dropsondes below 700 hPa are ~ 5–10% smaller than Control's; these differences are statistically significant for some layers.