Toward assimilation of cloud radar data for improvements in mesoscale forecasts

Unveiling the complex links between relevant physical processes at work in the Earth's atmosphere is a step toward understanding the weather and climate of Earth. Processes relating to the atmospheric branch of the hydrological cycle play a critical role in climate change through their impact on the atmospheric energy budget. Clouds, and in particular ice clouds, are an essential component of the atmospheric water cycle about which much uncertainty still exists. The paucity of observations and the inherent difficulties in modeling ice clouds both contribute to this uncertainty. Yet, the impact of this type of clouds on the atmospheric energy budgets through their influence on radiative and latent heat processes, especially in the upper-troposphere, is considerable. In recent years, much progress has been made in the description of clouds in numerical weather prediction models. Together with advancements in instrumentation and cloud observing capabilities, such as the ones afforded by short wavelength cloud radars, this has granted a better general understanding of ice cloud formation and evolution in relation to the ambient conditions. Improving data sets and models is a crucial part of a strategy to produce a more accurate representation of ice cloud processes and consequent improvements in weather forecast and climate predictions. In particular, the synergy of both these elements---models and observations---is the most promising way to address the open question about the role of ice clouds in climate change. Data assimilation offers an elegant mathematical framework to bring observations and model forecasts together. While it is no substitute for model improvement and development, it represents a powerful tool to enhance the potential of the observations through the use of the model, and to improve the model performance providing optimal model initialization through the use of measurements. In this larger picture, the focus of this research is to address some specific questions related to the combined use of model and cloud radar observations in advancing modeling and prediction of cirrus clouds. The core of this work is the development of an assimilation system based on variational principles to perform both sensitivity studies and assimilation experiments with synthetic and real radar reflectivity measurements. To this end, an ice growth model and its adjoint are derived and used in one dimensional time-dependent variational assimilation studies. Sensitivity studies are performed and key parameters in cirrus cloud physics are identified. A more complex model---the Regional Atmospheric Modeling System---is also used in cirrus simulations and its skill assessed with the goal of using this model in four-dimensional variational experiments involving the use of radar data. Results are promising and show both the feasibility and the great potential of incorporating cloud radar observations into mesoscale models to improve cloud prediction.