Study of Atmospheric Turbulence in Urban Areas

Wind turbulence models designed for large aircrafts are not suitable for small aerial drones flying slow and low. Small drones are more sensitive to turbulence than large aircrafts, and fly in a flow field that is strongly affected by topographic and thermodynamic variations of the underlying terrain.

The goal of this research project is to advance the current understanding of near-surface turbulence in urban environments, and to derive improved wind turbulence models for the safe design and certification of aerial vehicles.

Funding: Amazon.com Inc. - Prime Air Program

Non-Equilibrium Boundary-Layer Flow Over Rough Surfaces

Non-equilibrium turbulence is the rule in geophysical and engineering flows. Fundamental questions remain unanswered regarding the structural changes of turbulence under non-equilibrium conditions, challenging our ability to comprehend and model flow phenomena across a wide range of scenarios.

This research project aims at elucidating fundamental mechanisms responsible for momentum and kinetic energy transport in turbulent flow over rough surfaces under non-equilibrium conditions induced by external pressure gradients.

Hurricane Boundary-Layer Turbulence

Hurricanes account for a significant portion of damage, injury, and loss of life that is attributed to natural hazards and are the costliest natural catastrophes in the US. Yet, considerable uncertainties remain on the air-sea interaction process, which in turn affects wind loads on off-shore and on-shore structures.

This research project leverages direct and large-eddy simulations along with measurements to elucidate fundamental mechanisms supporting air-sea exchanges in hurricane boundary layer turbulence.

Funding: Computing Research Association CIF2020-CU-64

Data Driven Models for Vegetation-Atmosphere Interaction

Land surface models used in climate simulations to describe exchange processes between plant canopies and the atmosphere are often based on simplistic phenomenological assumptions and their parameters are calibrated manually. This project will develop a physics-informed data driven surrogate model that accurately and efficiently describes exchange processes between plant canopies and the atmosphere under a range of realistic ambient conditions.

The surrogate model will be trained using results from high-fidelity simulations of airflow-vegetation interaction and its parameters will be calibrated (for a specific site) making use of Bayesian inference and available in-situ and remote-sensing measurements.

Funding: LEAP Center, Columbia University

Spatial Reconstruction of Turbulence via Machine Learning

Machine learning provides an attractive complement to classical physics- and math-based methods for forward and inverse problems in turbulence, given its inherent ability to learn non-linear relations between variables directly from measurements.

This research project combines machine learning with direct numerical simulations to reconstruct 3-D turbulent flow fields from sparse and possibly corrupt measurements thereof.

Funding: Data Science Institute, Columbia University

Snow Transport in Katabatic Winds

An understanding of the surface mass balance of the ice sheets is critical for predicting climate change, future sea level rise, and for interpreting ice core records. Yet, the evolution of the ice sheets through snow deposition, erosion, transport, and sublimation in katabatic winds (which are persistent across much of the Antarctic) remains poorly understood.

The goal of this project is to advance the present understanding of these processes leveraging direct numerical simulations and in-situ measurements in Antarctica.

Funding: National Science Foundation NSF-OPP-2035078

Towards a Mechanistic Epidemiological Modeling Framework

Existing epidemiological models are based on simplistic relations to determine how individuals contract an infection, and do not account for microscale processes (droplets and aerosol dispersion) delineating virus transmission opportunities.

The goal of this project is to bridge this knowledge gap by formulating the first individual-based, mechanistic epidemiological model, which captures the dynamics of pathogen-laden droplet dispersion and aerosolization under a range of ambient conditions and interactions between individuals.

Funding: Office of the Dean, The Fu Foundation School of Engineering and Applied Science, Columbia University