Students share HDRFS research highlights at annual meeting

Students share HDRFS research  highlights at annual meeting

The dozens of poster presentations by early career scientists included new work alongside updates to ongoing projects.

By Amanda Heidt

May 2025

Students presented their work during a networking and poster session at the 2025 HDRFS Annual Meeting. (Credit: Mayara Cueto-Diaz)

This year’s HDRFS meeting, held April 6-8 in Reno, Nevada, brought the entire team together for a third time to highlight the progress made towards the program’s goals over the last year. In addition to updates from component leads, attendees heard from invited speakers whose research aligns with the Harnessing the Data Revolution for Fire Science project and visited the main study site, located along Red Rocks Road roughly 45 minutes away from the University of Nevada, Reno (UNR) campus.

On the second afternoon, students affiliated with the HDRFS project walked researchers through their work during a networking and poster session. The nearly two dozen presentations spanned all six HDRFS components and included completely new projects alongside updates to those first shared during the 2024 meeting in Las Vegas.

Several students presented research focused on making emerging technologies more accessible. Deep learning models (DLMs) are increasingly in-demand, for example, but can be costly to train and are not always useful for on-board processing of data collected in the field. Khondker Fariha Hossain, a PhD student in the lab of UNR computer scientist Alireza Tavakkoli, shared work aimed at taking the power of larger, more powerful DLMs and, through a process called distillation, transferring their knowledge to so-called lightweight models. The framework, called Teach-Former, can analyse data from satellite imagery, drone footage, or LiDAR directly on laptops, sensors, or drones with far less computation power.

Another PhD candidate in Tavakkoli’s lab, Gunner Stone, likewise shared a project called LiDAR Toolkit that allows users to create synthetic LiDAR simulations that are more cost-effective for teaching and research. Because the platform isn’t limited by the same constraints as a true LiDAR scanner, it can be configured in novel ways, such as by stacking scanners or by making a point cloud that mirrors an aerial pass over a site to map its vegetation. These scans can in turn be used to train deep learning and natural processing models, bringing down the costs associated with creating them.

“We’re now able to augment the data sets we already have of real tree or shrub scans with these fake scans where we have ground truth data,” Stone says, adding that such strategies are used by many artificial intelligence companies today. “Data can be expensive, and I’ve had a lot of fun recreating features that may previously have been locked behind a proprietary paywall.”

Emily Christensen, graduate student at UNLV, presenting her research poster on “Assessing the Effect of Slope and Aspect on Burn Severity for the Dixie Fire in California”. (Credit: Mayara Cueto-Diaz)

Other students continued to build out the team’s understanding of the factors that influence burn severity in sagebrush ecosystems, collecting data from fires across California, Nevada, and Colorado. Emily Christensen, a PhD student in the lab of Haroon Sahotra, a hydrologist at the University of Nevada, Las Vegas, presented work from the 2021 Dixie Fire, the largest and most destructive fire in California that year. She found that increasing slope leads to more severe fires, likely because of the role that slope plays in driving wind, while aspect did not seem to influence severity to the same degree. Across the aisle, work by other students in Sahotra’s group linked burn severity to plant ecosystem recovery—all research that will be probed in more depth at the field site along Red Rock Road in the coming year.

“These wildfires can be so large and unpredictable, so it will be so helpful to test some of these hypotheses in a more controlled setting at the site,” Christensen says. “We’re already collecting pre-burn data out there, so we’re set up really well for what’s to come.”

Lastly, students associated with the Education and Workforce Development (E-WFD) component shared updates on the group’s efforts to expand HDRFS findings into new spaces.

Grace Frazee, undergraduate student at UNR, presenting her poster “Growth of NevadaTeach Pre-Service Teachers”. (Credit: Mayara Cueto-Diaz)

Grace Frazee, an undergraduate at UNR, has spent the last several months interning through a partnership between HDRFS and the university’s teaching program, NevadaTeach. In her poster presentation, Frazee described her experiences traveling to six schools across Washoe County to teach middle and high school students the basics of fire science. The learning modules, developed in collaboration with scientists at the Desert Research Institute, touch on topics like soil science, fire ecology, biodiversity, and invasive species management, often using games and other interactive elements to introduce students to complex topics.

The program, Frazee says, has not only brought HDRFS research to younger audiences—hopefully inspiring future fire scientists—but helped her cultivate confidence as well, such that she now hopes to one day teach mathematics.

“At first I wasn’t sure if I was even going to be able to teach, just because I’ve always been really quiet,” she says. “But after doing the internship, it’s clear that I can, and even the challenging classes have started to feel like something I’m capable of doing. This internship has really changed things for me.”