Hello and welcome to my homepage.

My undergraduate education was a mix of data science, mechatronic engineering, and management science, but I realised I wasn’t really interested in that. I spent a lot of time learning about data science, computational science and simulation techniques and used them to conduct research in the field of atmospheric science. I am currently conducting research as a visiting student in the Atmospheric Environment Research Lab at Westlake University. In my work here, I strive to integrate simulations based on physical models and fluid dynamics with insights derived from real-world observational data. My goal is to enhance both aspects and to design interpretable machine learning models and simulations that are more aligned with practical realities.

This work is part of the work of the research group and we aim to explore many interesting questions:

1. How can irregular data from unconventional sources, such as cars and aircraft, be used for kilometre or even sub-kilometre scale simulations and inversions of the Earth system, e.g. to locally improve the resolution of large-scale models using vehicle-based data? How can there be appropriate coupling between different classes of observations? Can vehicle-based data be used to obtain regional statistical characteristics of emissions of various GHGs and to complement emission inventories?
2. How can these data be used for data assimilation across scales? For example, how can these data be assimilated in a suitable way into existing models designed for satellites and fixed base stations, and how should their uncertainties be quantified?
3. How to quantify the impacts of buildings with high emissions on the surrounding environment, e.g., do plume under extreme conditions have an impact on human health?
4. Could higher resolution observations have unintended effects, e.g. for new scientific discoveries, or for training specific macromodels/simulation models? Could more small-scale source discovery and emission inventorying be done directly?
5. Will different ecological environments, e.g., cities, farmland, wetlands, have different emission patterns, not only in terms of source share, but also in terms of spatial and modal distribution of sources? And how can this information be used in Earth system modelling?

I am actively seeking opportunities to pursue a PhD. I have a particular interest in the intersection of data science and atmospheric science. On one hand, data science methods such as machine learning have been applied to fundamental and critical issues in atmospheric science, including computational fluid dynamics and numerical weather forecasting. On the other hand, techniques like data assimilation are helping us to re-evaluate essential components of data science, such as machine learning and deep learning, which are grounded in Bayesian theory. I’m sure I’ll find something very interesting here!

My work and research

A more detailed account can be found in my CV.

Works

• Designed a UAV control algorithm for conducting searches for sources of pollutant gas emissions
• Short-lived air pollutant prediction using exploratory data analysis, machine learning and deep learning
• Exploring transformation patterns among atmospheric pollutants in urban environments using Wavelets Methods
• Analysing regional methane emissions and patterns using vehicle data

Interests

• Data Science, Signal Processing, Simulation and Soft Computing
• AI4Science, including Scientific Simulations and Scientific Discovery
• Earth system science, e.g., impacts of climate change on human health, economic activities and cultural heritage
• Embedded system, like Single-board Computer and Microcontroller Unit

My background and history

The first research group I joined focused on developing control algorithms for application in UAVs, as you can ascertain from my publication record. Given the extensive use of UAV technology in warfare, I have decided to withdraw from this field.