Abstract
A simple adaptive thinning methodology for Atmospheric Infrared Sounder (AIRS) radiances is evaluated through a combination of observing system experiments (OSEs) and adjoint methodologies. The OSEs are performed with the NASA Goddard Earth Observing System (GEOS, version 5) data assimilation and forecast model. In addition, the adjoint-based forecast sensitivity observation impact technique is applied to assess fractional contributions of sensors in different thinning configurations. The adaptive strategy uses a denser AIRS coverage in a moving domain centered around tropical cyclones (TCs) but sparser everywhere else. The OSEs consist of two sets of data assimilation runs that cover the period from 1 September to 10 November 2014, with the first 20 days discarded for spinup. Both sets assimilate all conventional and satellite observations used operationally. In addition, one ingests clear-sky AIRS radiances, the other cloud-cleared radiances, each comprising multiple thinning strategies. Daily 7-day forecasts are initialized from all these analyses and evaluated with a focus on TCs over the Atlantic and Pacific. Evidence is provided on the effectiveness of this simple TC-centered adaptive radiance thinning strategy, in full agreement with previous theoretical studies. Specifically, global skill increases, and tropical cyclone representation is substantially improved. The improvement is particularly strong when cloud-cleared radiances are assimilated. Finally, the article suggests that cloud-cleared radiances, if thinned more aggressively than the currently used clear-sky radiances, could successfully replace them with large improvements in TC forecasting and no loss of global skill.
AIRS impact on the analysis and forecast track of tropical cyclone Nargis in a global data assimilation and forecasting system
