June 28, 2024
Student Project Highlight: Remote Sensing Assessments of Social Trails Fragmenting Subalpine Paradise Meadows
By Kylie Baker
As summer approaches, Mount Rainier National Park, located about two hours south of Seattle, transforms into a popular destination for hikers and outdoor enthusiasts. On the western side of the park, the area known as Paradise is particularly beloved for its subalpine meadows, the Jackson Visitor Center, the historic Paradise Inn, and breathtaking mountain views. However, while exploring the natural beauty of this iconic spot is beneficial to human well-being, it is equally important to protect the mountain’s delicate ecosystems.
Social trails significantly threaten the environment by damaging plant ecosystems with short growing seasons. They lead to the degradation of meadow habitats already at risk due to climate change and tree encroachment caused by reduced snowpack, and they harm the species that depend on these ecosystems. Increased park visitation exacerbates this issue, as more people means more off-trail wandering and more social trails.
Through the Pacific Northwest Cooperative Ecosystem Studies Unit (PNW CESU) and in collaboration with the National Park Service (NPS) Ecologist Beth Fallon, University of Washington researchers Dr. Meghan Halabisky, Dr. Monika Moskal, and research assistant Lindsey Skidmore are addressing this problem through their project “Remote Sensing Assessments of Social Trails Fragmenting Subalpine Paradise Meadows” (Project ID: P23AC01208). They aim to map social trails in Paradise through the NPS field data and high-resolution imagery analyzed with machine learning.
During the summer months, the NPS interns Frannie Nelson and Mitra Aflatooni collected field data, noting disturbed sites and mapping their locations and severity. They were supported by the NPS Scientists in Parks and Washington National Park Fund (WNPF) funding, and their work designing data collection and field data made this research possible. Frannie was also a NPS seasonal employee last year and provided significant reporting and data management support for the work.
The University of Washington researchers then tested 21 different remote sensing inputs, including a Digital Terrain Model (DTM) and a Normalized Difference Vegetation Index (NDVI), but found that high-resolution aerial imagery was the most crucial for detecting social trails. They combined the NPS-collected field data of known impacted areas with high-resolution imagery from 2021—one from July and one from September—at resolutions of 25 and 60 centimeters. The images from the summer months were utilized as the lack of snow allowed social trails to be seen more clearly. Using the field data and high-resolution imagery, they created a training dataset with about 3,000 points of impacted and natural areas.
The training data and imagery were run through a Random Forest Algorithm, resulting in a prediction model with an output of social trail likelihood from 0 to 1. The model achieved an overall accuracy of 81.4%, though detecting tiny social trails was challenging due to snow, trees, shadows, and varying image resolution. The model also struggled to distinguish between human-impacted areas and those affected by natural erosion.
As meadow extent decreases, monitoring these ecosystems becomes even more critical to ensure their survival. The most current map created by the researchers has been provided to the NPS for management purposes and will continue to be refined. The next steps involve collaborating with the NPS this summer to collect additional data during snow-free months and validate the existing model. This involves going into the field to the predicted points, marking them, and then analyzing the accuracy of the model in predicting social trail points. The team also hopes to incorporate more high-resolution imagery throughout the seasons, although acquiring such imagery is costly and difficult due to Mount Rainier’s often cloudy weather.
Lindsey Skidmore will continue working on this project as an intern with the NPS through Scientists in Parks, collecting more field data and validating and improving the model. The team plans to implement a new field monitoring plan aimed at achieving higher accuracy with less intensive data collection. Through these efforts, they hope to better protect the fragile ecosystems of Paradise and ensure that future generations can enjoy its beauty without causing harm.