Sensitivity of Tropical Cyclone Forecasts to the Loss of Low Earth Orbit Satellite Observations

Abstract Tropical cyclones are among the most destructive natural hazards, impacting millions of people worldwide each year. Accurate and timely forecasts are therefore essential for effective preparedness and risk mitigation. Forecast skill depends on both the performance of numerical weather prediction models and the observations assimilated into the system to provide the initial conditions. In this study, we assess the impact of different observing systems, with particular emphasis on microwave (MW) and infrared (IR) measurements from Low Earth Orbit (LEO) satellites, on tropical cyclone forecasts. We evaluated predictions for several historical tropical cyclones using NASA's Goddard Earth Observing System (GEOS) data assimilation framework. The results indicate that excluding MW and IR satellite observations from the assimilation process leads to a substantial degradation of track forecasts. The forecasts for maximum wind speed and minimum sea level pressure are comparatively less sensitive to denial of MW and IR observations and the impact overall diminishes at longer lead times. In addition, increasing the model horizontal resolution from approximately 25–12 km leads to notable improvements in the intensity forecasts but some limited impro vements in track predictions.