Ph.D. student studying cloud physics and atmospheric turbulence at the University of Utah. Research focuses on scale invariance, multifractal analysis, and emergent laws in weather and climate systems.
Education and Awards
Ph.D. Atmospheric Sciences (expected 2026), M.S. Atmospheric Science (2023), B.S. Physics (2020). Norihiko Fukuta Memorial Award for Outstanding Graduate Student Publication (2024). ACP and NPG Highlight papers (2024).
Research Publications
Global sonde datasets do not support a mesoscale transition in the turbulent energy cascade - Atmospheric flow is commonly thought to be controlled by several distinct physical mechanisms, each operating at a unique spatial scale. Using global radiosonde data and dropsonde programs, we computed structure functions of horizontal wind across vertical separations of 0.2-8 km and horizontal separations of 200-20,000 km. We found no spectral "transition" regime; instead, observations support a lesser-known anisotropic turbulence theory where larger, flatter circulations continuously deform into more circular ones at smaller scales. This was the first independent test of the Lovejoy-Schertzer anisotropic cascade theory proposed 40 years ago.
Climatologically invariant scale invariance seen in distributions of cloud horizontal sizes - Cloud size distributions follow power laws across more than five orders of magnitude, from small cumulus to large organized systems. We found remarkably stable scaling exponents across different seasons, latitude bands, ocean versus land, and multiple satellite platforms including GOES, Himawari, MODIS, and EPIC. This climatological invariance suggests cloud sizes are controlled by fundamental stability properties rather than surface temperature, aerosols, or Coriolis effects. The universality implies that simulating the statistics of the largest clouds may suffice to constrain the full distribution via scale invariance.
Finite domains cause bias in measured and modeled distributions of cloud sizes - Decades of disagreement in cloud-size scaling exponents trace primarily to finite-domain truncation artifacts, not differences in statistical estimators. We showed that both including and excluding domain-truncated clouds introduces systematic bias, and that apparent "exponential tails" and scale breaks at large sizes are often artifacts of domain boundaries. We developed a correction procedure based on filtering bins where truncated clouds dominate, which yields stable exponents across domain sizes. This methodology applies broadly to any geometric size distribution analysis.
Toward less subjective metrics for quantifying the shape and organization of clouds - We propose two distinct fractal dimension metrics with clear physical interpretations, exposing pitfalls in decades of cloud fractal analysis. The common perimeter-area method fails to capture the true fractal dimension of clouds. We show that holes in clouds must be filled before computing individual fractal dimensions, and introduce an ensemble dimension better suited to satellite observations. These objective, physics-linked metrics can validate cloud organization in global cloud-resolving models.
Software
scaleinvariance - Python package for multifractal simulation and Hurst exponent analysis
objscale - Fractal dimension and size distribution analysis for binary arrays