Investigating Spire GNSS RO Bending Angle Assimilation Impacts on HWRF-forecast Hurricane Earl (2022) Structural and Intensity Changes: A Case Study

Hurricane Earl (2022) presents an interesting case study from a basic scientific and a forecast evaluation perspective. The storm intensified from its precursor disturbance to a 90-kt hurricane over the 2-7 September period despite experiencing unfavorable moderate-to-strong vertical wind shear (VWS). During this time it moved west-northwestward north of the Lesser Antilles and then recurved northward into the subtropical western North Atlantic. Then, on 8 September, despite the environmental VWS relaxing, Earl weakened temporarily before undergoing a final intensification burst on 9 September to reach a final 95-kt peak Category 2 intensity several hundred miles southeast of Nova Scotia. National Hurricane Center official forecasts predicted Earl would reach Category 4 intensity over this period, owing to an over-intensification bias in Numerical Weather Prediction (NWP) forecast model guidance.

This study explores the sensitivity of Earl’s 8 September intensification pause and subsequent re-intensification to the lower-to-middle tropospheric vortex moisture fields several days prior. This is done using output from a set of offline Hurricane Weather Research and Forecasting (HWRF) model observing system experiments investigating Global Navigation Satellite System (GNSS) radio occultation (RO) data assimilation (DA) impacts on several 2022 Atlantic hurricanes. GNSS RO signals received by low earth orbit (LEO) receiver satellites are processed to retrieve RO ray bending angles, which yield information about atmospheric water vapor, temperature, and pressure profiles along the RO ray limb soundings. Spire Global, Inc. provided the National Oceanic and Atmospheric Administration (NOAA) with ~5,500 daily RO profiles for NWP model DA as part of their 2022 Delivery Order-4 (DO-4) contract. Two sets of HWRF experiments have been carried out. The HWRF C2 experiment assimilates RO observations from government-funded missions, including the US-Taiwan COSMIC-2, while the C2SPIRE experiment also assimilates the DO-4 Spire RO data.

Here, we compare Hurricane Earl’s intensity and structure from C2 and C2SPIRE HWRF forecasts initialized on 12 UTC 05 September, following a 60-hour DA cycling period with observations assimilated every 6 hours. Spire RO bending angles assimilated over the prior eleven cycles had an overall drying impact on the Earl analysis vortex, reducing relative humidity (RH) by 5-10% in some areas below 850 hPa. Interestingly, both the C2 and C2SPIRE intensity forecasts are similar through 00 UTC on 08 September, and they capture Earl’s intensification rate and asymmetric convective structure while the storm is experiencing moderate-to-strong VWS. Thereafter, C2 fails to forecast Earl’s temporary weakening and forecasts a large +12 m s^-1 maximum surface wind (V_MAX) bias at peak intensity. C2SPIRE, on the other hand, captures some of the observed 8 September temporary weakening, leading to a smaller +7 m s^-1 V_MAXbias. Comparing the structural evolution of the C2- and C2SPIRE-forecast Earl against in-situ and satellite observations leads us to hypothesize that the drier C2SPIRE analysis vortex reduced the HWRF over-intensification bias by either (i) rendering Earl’s inner core deep convection less resistant to a dry air intrusion that likely caused the 8 September temporary weakening and/or (ii) reducing the “active” eyewall convection-facilitated main outflow intensification when Earl began interacting with an upper-level midlatitude trough late in the forecast period. These results further suggest that despite the Spire mission’s reduced spatial coverage over the tropics compared to COSMIC-2, more RO observations can still provide additional value to NWP model TC forecasts by helping to correct model-specific RH forecast biases. However, more case studies are needed to confirm our findings.