Sing-Chun (Sally) Wang

About me

"Give me a lever and a place to stand on, and I will move the Earth." --- Archimedes. After learning the knowledge of various subjects after I got my bachelor degree, I think Atmopsheric Science is my ideal end. Nowadays, the impact of global warming is more and more serious. It not only makes the sea level rise but also increases risks of some natural disasters like fire, drought, and floods. Those changes have large threats to wildlife besides human beings. Therefore, I want to be the indispensable fulcrum wherever we can solve these problems and make the Earth better.

Biography

Sing-Chun (Sally) Wang, Ph.D., obtained her bachelar of Biology and minor in Environmental Engineering and Earth Science from National Cheng Kung University (NCKU), Taiwan. After a field trip of a polluted Erren River in Tainan, Taiwan, the real scenes were so shocking that she started to focus and care about the environmental problems. Therefore, her strong interests in Enverionmental Sciences motivated her studies in Atmospheric Sciences.

She obtained her doctoral degree in Atmospheric Science at University of Houston (UH), USA in May 2019. During her PhD study, she used GEOS-Chem model, satellite observations, and ground observations to study the transport patterns of Central American fires and to quantify the impact of the fires on Gulf Coast air quality. She also developed a machine learning model to predict burned area in the South Central US. The developed model is able to obtain higher prediction accuracy and to provide better understanding of wildfire mechanisms in the South Central US.

From March 2020, Sally started working as a postdoctoral fellow and she is now a data scientist at Pacific Northwest National Laboratory and her on-going research is to utilize explainable artificial intellegence (AI) to better predict and understand wildfires.

You can find my complete curriculum vitae in PDF format here and my google scholar page can be accessed here.

Research

Research interests

The research goal of my studies is to better understand wildfire behavior and its impact. An increasing trend of intense wildfires associated with severe drought under a warming climate has been reported by many studies. Wildfires have led to poor air quality by emitting large amount of particles and ozone precursor gases. Those pollutants can modify cloud properties, affect energy balance, and further influence weather and climate. Thus, I am mainly interested in the interactions among wildfires, air pollution, weather, and climate.

Fire smokes from Central America to US

Fire emissions from Mexico and Central America are transported regularly to the U.S. Gulf Coast every spring under prevailing circulation patterns and affect U.S. air quality. We used a GEOS-Chem passive tracer simulation to develop the climatology of transport pathways of fire emissions over a long-term time period of April and May, 2002-2015 and estimate their adverse air quality effects for urban areas along the Gulf Coast. Considering both transport and fire emissions, we identified approximately 9% of the study period (59-88 days of 854 days) were influenced by the Central American fire emissions. The impacts on surface air quality were found at several major urban centers along the Gulf Coast, including Houston and Corpus Christi in TX, New Orleans in LA, Mobile in AL, and Pensacola in FL. Compared to clean maritime flow from the Gulf of Mexico, these events were estimated to result in average enhancements of maximum daily average 8-hr (MDA8) ozone and daily PM2.5 in the Gulf Coast cities of 3-12 ppbv and 2-5 microgram/m3, respectively.

The details of this study and results can be found in Wang et al. (2018) JGR.

Wildfire model for South Central US

One common method to explain and investigate the relationship between wildfires and factors is regression, while most data-driven regression methods would favor the majority class (here referring to unburned or small-burned grids) in the unevenly-distributed data (wildfire burned area data), which would lead to underestimation of large fires. To solve this, we integrated multiple machine learning algorithms to develop a prediction model of 0.5 deg x0.5 deg -gridded monthly wildfire burned area over the South Central United States during 2002-2015 and then use this model to identify the relative importance of the environmental drivers on the burned area for both the winter-spring and summer fire seasons of that region.

The details of this model and results can be found in Wang et al. (2020) ACP.

A universal wildfire model for the continental US

Based on the knowledge and experience of wildfire models for the South Central US, I'm currently developing a machine learning model to estimate burned area of the Continental US. By utilizing the novel interpretation method, we're able to open the black box of machine learning model to better understand the leading factors controlling burned area for different regions of the US for future management, prevention/mitigation, and projection.