Research

Projects

ARCO-Mars

I developed a cloud-optimized archive of three SOTA Mars reanalysis datasets: EMARS, OpenMARS and MACDA. Reanalysis refers to the retrospective sequence of analysis steps that combines GCM forecasts with observations through data assimilation. The datasets are accessible through Hugging Face and Earth Mover in Zarr format.

It provides wider access of Mars weather and climate data, enabling applications in space science, machine learning and mission engineering.

FVCOM Lake Coupling with NOAA UFS for Great Lakes

As a Postdoctoral Research Fellow at University of Michigan, I work on technical development of NOAA’s operational forecasting system: Unified Forecast System (UFS). The Great Lakes hold one of the world’s largest freshwater resources, and lake–atmosphere interactions drive substantial annual precipitation across the region. Great Lakes Operational Forecasting System (GLOFS) based on FVCOM-CICE lake model is coupled to UFS configured to NOAA’s next generation operational forecasting model Rapid Refresh Forecast System (RRFSv1). Asynchronous, one-way coupling of three-dimensional lake hydrodynamics improves the representation of Great Lakes for short-range forecasting models.

3-D Animation of Cloud Formation During Lake-Effect Storm: UFS-FVCOM Simulation for RRFSv1 Configuration

Post-Training Quantization of AI Weather Models

Using NVIDIA Earth2Studio, I implemented post-training quantization (PTQ) on weight and activation bit representation within the AI weather models. Evaluation of quantized models against standard forecast metrics show promising performance at a forecast horizon of 60 hours. This proof of concept demonstrates PTQ in multi-step deep learning models for geophysical fluid dynamics applications. Here’s the GitHub.

Evaluation of Physical Consistency in AI Weather Models

I also participated in Dynamical Core Model Intercomparison Project, and I have been working with AI emulators for atmospheric dynamics. Checkout by blog on creating a simple workflow for AI driven weather forecast.

Community Science and Sustainable Development

As a AGU Community Science Fellow, I managed a 18 month long project focused on community driven data collection to measure underwater acoustic signatures in Calcasieu river. The project identified bioacoustic signatures and detected changes relative to ship traffic. In collaboration with Habitat Recovery Project, the project aimed at understanding the impact on underwater noise pollution for sustainable fisheries in Louisiana Project Page.

I am serving as a Community Scientist with Boynton Beach to investigate urban heat island effect in Florida and develop strategies for urban heat resilience Project Page.

Microwave Remote Sensing of Jupiter with NASA Juno Mission

As a scientific collaborator with NASA Juno mission, I worked on scientific analysis of Juno Microwave Radiometer data. My investigation of Jupiter’s deep atmosphere revealed highly depleted alkali metals, and their thermochemistry forming thin alkali sulphide and halide clouds. Heavy elemnent inventory in a gas giant has implications on its formation and evolution.

Elemental Abundances in Jupiter’s atmosphere. Alkali metals are expected to depleted in Jupiter’s atmosphere

Space Environment of Planets and Satellites

My analysis of Jupiter’s polar atmosphere observations at microwave frequencies provided evidence of high-energy electron precipitation and their time-dependent behavior. I am also collaborating with Juno radio occultation team to understand the auroral ionosphere. I received a NASA Group Achievement Award for Juno Radio Occultation.

I am actively working on investigating the plasma interactions in lunar orbit and near-surface environment.