Areas of Expertise

We bring decades of expertise in development of global Earth system models and advanced techniques for assimilation of satellite measurements into these models. Our approaches are applied to weather and air-quality forecasting, historical reanalyses, and observing system simulation experiments (OSSEs) that support future space missions. Our experience encompasses enhancement of aerosol representation within the Global Earth Observing System (GEOS) model, implementation of Ensemble Kalman Filter (EnKF) algorithms for assimilating aerosol observations, and use of machine learning algorithms to improve aerosol retrievals.

We develop advanced algorithms for assimilating space-borne cloud observations into global Earth system models. Our approach leverages Bayesian inference to estimate sub-grid probability distribution functions (PDFs) of water vapor and cloud condensate, utilizing high-resolution satellite data to enhance model accuracy and representation.

A model-based Observing System Simulation Experiment (OSSE) provides a structured framework for numerical experimentation, enabling the simulation of observations derived from Earth system model fields complete with a parameterized representation of errors. Our expertise lies in applying a rigorous OSSE methodology to the measurement of weather, clouds, aerosols, and trace gases, supporting both current and future satellite observing missions at the global kilometer scale.

Our expertise centers on the development of high-performance, flexible software infrastructure designed to build and integrate weather, climate, data assimilation, and related Earth science applications. We focus on advancing the Earth System Modeling Framework (ESMF) by creating integrated development environments and refining usability layers (such as MAPL and NUOPC) to enhance interoperability. Additionally, we develop advanced data assimilation algorithms leveraging the Joint Effort for Data assimilation Integration (JEDI) framework, ensuring robust and adaptable solutions for the Earth science community.

We specialize in the development of advanced Bayesian inference algorithms for sub-pixel fire characterization, enabling precise quantification of wildfire energetics, combustion phase dynamics, and their atmospheric impacts using satellite observations. These algorithms form the foundation for estimating gridded biomass burning emissions that are essential for driving global Earth system models. Our approach emphasizes top-down methodologies, leveraging sub-pixel fire property algorithms that integrate data assimilation and machine learning techniques for enhanced accuracy and insight.

We have demonstrated expertise in the development of novel retrieval and atmospheric correction algorithms for multi- and hyperspectral satellite data that have enabled the production of multiple long-term data records used extensively by the biospheric community. Our work includes innovative algorithms for solar-induced chlorophyll fluorescence retrieval, machine-learning-based upscaling of ground measurements to estimate gross primary production (GPP), and retrievals through clouds.

We bring decades of experience in the development of advanced algorithms for the retrieval of atmospheric trace-gas concentrations, including ozone (O₃), sulfur dioxide (SO₂), and nitrogen dioxide (NO₂), from passive hyper-spectral satellite observations. Our work has pioneered novel methodologies, such as cloud-slicing techniques for deriving vertical trace-gas profile information and machine-learning-based approaches for noise mitigation. These advances have enabled the generation of high-quality, long-term, multi-instrument satellite data records that are widely utilized by the atmospheric composition research community.

We possess extensive experience in the formulation of satellite instruments and missions, providing science guidance from initial concept through the definition of scientific requirements and the development of compelling proposals

We leverage advanced artificial intelligence and machine learning algorithms extensively to enhance both modeling and remote sensing applications. This approach enables us to extract deeper insights, improve predictive capabilities, and drive innovation across a wide range of Earth science challenges.