The Impact of All-Sky Hyperspectral Infrared Radiance Assimilation on the Simulation and Forecast of Hurricane Sally
in GEOS


Wei Gu (1,2), Yanqiu Zhu  (1), Yingtao Ma(3), Oreste Reale (1,2), Erica McGrath-Spangler (1,4), Niama Boukachaba (1,4), Manisha Ganeshan (5,4)
Ata Akbari Asanjan (6)

1 GMAO @ NASA/GSFC
2 Science Systems and Applications, Inc.
3 CIRA @ NOAA/NESDIS/STAR
4 Morgan State University
5 Climate and Radiation Laboratory @ NASA/GSFC
6 USRA RIACS @ NASA/Ames Research Center


The hyperspectral IR radiance observations have been one of the major data sources in the data assimilation system, including GEOS.
However, they are underutilized for those peaking in the lower and mid-troposphere in clear-sky data assimilation. This is due to the
cloud detection procedure, which removes a significant portion of cloud-affected observations from the assimilation. To use these
cloud-affected observations, one approach is to assimilate them directly for all-sky data assimilation.

The framework for all-sky assimilation of hyperspectral IR radiance observations has been developed in GEOS. The capability to simulate cloud-
affected IR observations has been evaluated along with the corresponding sensitivites with respect to all hydrometeors. The symmetric
cloud effect has been used as a cloud proxy, providing a balanced representation that mitigates the discrepancies between observations
and model simulations for IR all-sky assimilation. The observation errors for these cloud-affected observations have been modeled as
cloud amount dependent and inter-channel correlated.

Although the all-sky framework has been tested in GEOS, the ultimate goal is to improve its impact
on the analysis and NWP forecast performance of GEOS. To achieve this, our initial focus is on improving the
simulation and forecast of hurricanes, specifically regarding intensity and track. Hurricane Sally, which formed between Andros
Island and Bimini in the Bahamas on September 11, 2020, at 18Z, will be used as a target case. The initial evaluation of the all-sky
observation operator shows that the simulated brightness temperature tends to have broader structures and lacks small-scale
details. This issue may be due to excessive clouds generated by the model or stronger-than-expected scattering from hydrometeors.
A variety of cloud overlap schemes and cloud LUTs, along with several other important aspects of all-sky IR radiance
observation assimilation, will be reassessed to address this issue and improve the impact on Hurricane Sally's simulation and
forecast.