CSE Community Seminar
March 14, 2025, 12-1PM
Conference Room 45-432 in Building 45
Bridging model and reality: data assimilation for turbulence reconstruction and debiasing simulations
Mengze Wang
Postdoctoral Associate,
Department of Mechanical Engineering, MIT
Abstract:
Numerical simulations of chaotic and multiscale systems, such as engineering turbulence and global climate, present significant challenges. Even at the highest level of simulation fidelity, due to the assumptions introduced in initial condition, boundary condition, or model parameters, numerical predictions often deviate from experimental measurements or field tests. Data assimilation provides a systematic framework to address these limitations by optimally infusing limited observations into numerical simulations, enabling the full access to all the scales. In this talk, I will first focus on canonical wall-bounded turbulence, where the incompressible Navier-Stokes equations are generally considered a perfect model for describing flow physics. Given limited velocity data, our objective is to construct a Navier-Stokes solution that reproduces these observations and predict the unknown flows. A critical data resolution is identified that ensures an accurate reconstruction of turbulence. Below this critical threshold, the origins of measurements become increasingly ambiguous, and only the flows within the domain of dependence of observations can be accurately decoded. I will then address the more challenging scenario of biased model equations, using climate simulations as an example. A prediction-correction strategy will be presented, where data assimilation is utilized to nudge the simulations towards observations. These aligned simulation-observation pairs are then used to train a machine-learning model that identifies and removes systematic biases. When applied to free-running climate simulations, this framework demonstrates significant improvements in capturing the statistics of rare events.
March 14, 2025, CSE Community Seminar
Mengze Wang
Postdoctoral Associate,
Department of Mechanical Engineering, MIT