Date of Award

January 2021

Document Type


Degree Name

Doctor of Philosophy (PhD)


Atmospheric Sciences

First Advisor

Jianglong Zhang


Observations from hyperspectral infrared sounder (HIS) instruments aboard earth-observing satellites have become a cornerstone of numerical weather prediction assimilation efforts – providing the largest decrease in forecast error of any assimilated satellite observations. The assimilation of infrared (IR) radiances is predicated on the assumption of clear-sky observations. Thus, any signal imparted upon the HIS radiances due to cloud or aerosol will likely result in unexpected and uncharacterized biases in analyzed temperature and humidity fields. Forecasts based upon these biased fields may have large inherent inaccuracies. The process of cloud and aerosol screening of passive satellite products and radiances is imperfect. Residual aerosol and cirrus clouds are found to contaminate HIS radiances assimilated from presumed clear-sky scenes at concerning rates (approximately 30% and 8% for the Naval Research Laboratory Variational Data Assimilation System, respectively). As such, the presence of an uncharacterized bias exists within model analyses.

To determine the biases a modified one-dimensional variational (1DVar) assimilation system is used for two studies: one for aerosol, one for cloud. For the aerosol study, observations of dust from the Island of Tenerife, Spain are used to create synthetic dust contaminated HIS observations. For the cloud study, a series of clouds of varying optical depth and cloud top altitude are simulated. Analysis biases greater than expected forecast uncertainties are found for both studies. Aerosol biases are smaller, likely due to lower thermal contrast with the lower atmosphere. For instance, at an average aerosol optical depth of 0.30 a peak temperature bias of 0.5 K and dew point bias of 1.0 K is found. Meanwhile, for cloud optical depths as small as 0.1, maximum temperature and dew point biases of 3 K and 10 K are shown.

Finally, a third study in similar vein to the first two simplifies the impact of aerosols on numerical weather prediction by examining the impact of aerosol optical model on broadband radiative properties. Observations above and within a dust aerosol plume collected during the Studies of Emissions and Atmospheric Composition, Clouds and Climate Coupling by Regional Surveys (SEAC4RS) field campaign are used to attempt radiative closure. Large variability for different commonly used aerosol optical models is shown for shortwave fluxes and heating rates of up to 50% and 400%, respectively. In the IR, variability is still relatively smaller, but still very large at 3% for flux and 25-50% for heating rates. Finally, it is determined that aerosol analyses from models are not sufficiently accurate to provide accurate fluxes or heating rates.