From a young age, Yushu Xia was keenly aware of the importance of environmental health—a passion sparked by her grandfather, a geology professor and author of popular science books who inspired her love for science and education.
Today, Xia combines these interests as an assistant research professor at Lamont-Doherty Earth Observatory, which is part of the Columbia Climate School. Xia’s niche: the intersection of soil and agricultural science with modeling and data science, all of which are essential in “truly understand[ing] how soils function,” she says.
In the following discussion, Xia talks about her growing roster of innovative research projects that seek to understand and improve soil health; the importance of community and student engagement; and ways to stay ahead in a changing scientific landscape.
How did you become interested in environmental science? And soil health specifically?
I was deeply inspired by my grandfather, a geology professor whose greatest passion was influencing the next generation and communicating science to broad audiences. He wrote many children’s books that introduced young readers to the wonders of geoscience. His example sparked my interest in academia and my desire to use research as a way to inspire others. From an early age, I was motivated to address environmental challenges and contribute to positive change in society.
I have always been passionate about soils, which reside at the center of many important themes. For example, better soil management can reduce environmental pollution. It could also be considered a nature-based solution because putting more carbon into the soils can reduce carbon dioxide in the atmosphere. Most importantly, improved soil health management can enhance agricultural productivity, which directly benefits farmers and ranchers.
Studying soil health is both fascinating and challenging. Today, soil health research involves more than just collecting field samples and running lab tests. To truly understand how soils function, it’s essential to connect field and lab observations with data science and modeling techniques to capture processes across different scales. I enjoy this complexity and find soil health research deeply relevant to today’s society.
What projects are you working on right now?
One project I’m currently working on is in collaboration with the Agricultural Model Intercomparison and Improvement Project, AgMIP, which connects me with a global network of agricultural modelers. We recently launched a project to improve the modeling of soil carbon and nitrogen dynamics in agricultural lands. This effort brings together major U.S. agricultural modeling groups to compile, standardize and harmonize datasets. Our goal is to reduce model uncertainty and improve predictive accuracy. The outcomes will benefit U.S. farmers and ranchers while also informing policymaking to support food security.
In another project, I’m collaborating with the National Soil Survey Center to improve the modeling of soil health in the U.S. We are currently exploring the use of AI and deep-learning algorithms to model soil health across space and time. The modeling results will be visualized on a community-based platform, which can hopefully provide actionable insights to improve agricultural and forest management and guide policy decisions, as well as inform the next steps for national soil survey efforts.
“No matter how advanced models and algorithms become, understanding soil, plants and ecosystems on the ground remains essential, so students should not be afraid of getting their hands dirty.”
Another ongoing project aims to improve the estimation of rangeland carbon and water dynamics. For this work, we’re collaborating with NGOs such as the National Audubon Society, World Wildlife Fund and the National Fish and Wildlife Foundation. We are leveraging RangeStar, a novel model I developed during my postdoc that integrates remote sensing with process-based modeling. In this project, we are analyzing data from several million acres of grassland to refine RangeStar and generate more accurate estimates of soil health and ecosystem productivity. At the same time, our collaborators are assessing biodiversity, water quality and vegetation health. This interdisciplinary approach allows us to explore the drivers of multiple ecosystem service outcomes and examine their connections with soil health.
I’m also working with local partners to engage students and stakeholders more directly. For example, we recently established a campus experimental site at the Lamont Sanctuary Forest. More than 30 students, from high school to graduate levels, have participated in sample collection, lab tests, data analysis and model building to evaluate forest and soil health. Within New York, I’m also collaborating with the Stone Barns Center for Food and Agriculture, Hudson Carbon and the Cary Institute. This past summer, my lab sent teams of students to field sites across New York to conduct sampling and engage with local farmers.
How do you balance all these research projects with teaching and mentoring?
Balancing multiple research projects alongside teaching and mentoring is certainly challenging for an early-career faculty member. I’m a very energetic person, and my passion for soils allows me to devote significant time to all aspects of my work. That said, because my research sits at the intersection of soil and data science, it naturally connects to my teaching and mentoring activities, which makes everything more seamless.
I teach a class called Agroecology. The topics include not only crop yield and ecosystem productivity, which are extremely important, but also ecosystem outcomes related to soil health, water quality and availability, animal health, and wildlife biodiversity. The goal is to provide students with a holistic understanding of agriculture so they can better evaluate and improve agricultural management and policy.
You recently won the Innovator Award from the Foundation for Food and Agriculture Research. Can you talk a bit about that project and what it means for you?
This award means a lot to me, especially as I’m only in my second year at Columbia. Being selected affirms that FFAR sees the value in the work I’m pursuing, and I look forward to engaging further with their exceptional network.
In this project, I will collaborate with a wide range of stakeholders, such as farmers, ranchers and policymakers. We’re collecting field and lab measurements, as well as publicly available data from remote sensing, soil surveys and network-based observations. Additionally, we’re extracting environmental covariates representing weather, vegetation and topographic features to build predictive models. The goals are to improve our mechanistic understanding of soil carbon and nitrogen dynamics and their controlling factors, improve models and tools, and create broader impacts through stakeholder engagement.
I’m particularly excited about integrating remote sensing and management-driven models into a more flexible system, and engaging with stakeholders in an iterative way by collecting their feedback to continuously improve our modeling system to support timely management decisions. We’re also going to provide interactive decision-support tools where farmers and ranchers can use a web or phone app to explore how different management and weather scenarios may affect outcomes. Students will be actively involved in this whole process for designing and implementing these platforms.
What advice do you give your students who are trying to enter this field?
My advice for students is to stay curious, proactive and ahead of the curve. It’s important to read widely and engage with researchers and stakeholders to understand what has already been done and identify new directions in research that can benefit society. For example, when I was doing my PhD, few students were exploring the intersection of soil and agricultural science with modeling and data science. With the rapid development of AI and new technologies, I recognized the potential of this area and positioned myself to take advantage of emerging opportunities. After graduation, I found many career opportunities in this direction. I encourage students to seek out similar niches where their curiosity and skills can have a meaningful impact and accelerate progress toward their career goals.
Looking ahead, I see a major hotspot in agricultural and soil research in improving precision agriculture, especially using sensor-based technologies like drones and remote sensing. This requires students to develop strong quantitative skills and embrace interdisciplinary approaches. At the same time, no matter how advanced models and algorithms become, understanding soil, plants and ecosystems on the ground remains essential, so students should not be afraid of getting their hands dirty. While much research remains theoretical, I believe applied work is equally important. Engaging with farmers, ranchers and policymakers allows researchers to incorporate practical feedback into iterative modeling systems, which will ensure that scientific insights translate into real-world impact.