Leveraging Data Science to Evaluate Vegetation Trends
CDS MS alumna Ningyuan (Teresa) Huang and CDS affiliated professor Sonali McDermid recently published “Moisture and temperature influences on nonlinear vegetation trends in Serengeti National Park”, in IOPscience’s Environmental Research Letters, a high-impact, open-access journal in the area of environmental science research. Using rigorous statistical approaches, their study revealed long-term, nonlinear vegetation trends in Serengeti National Park, which is a critical protected area in East Africa “where seasonal vegetation cycles support vast populations of grazing herbivores and a host of ecosystem dynamics”.
Previous research on the park demonstrates how non-climate drivers, such as land use, shape the park’s ecosystem. However, it remains unclear as to what extent changing climate conditions affect its vegetation trends. In their work, Teresa, Sonali and their co-authors, evaluate long-term changes in SNP leaf area index (LAI) that occurred from 1982 to 2016. (LAI is the ratio of one-sided leaf area per unit ground area.) The team assesses the LAI trends in relation to both temperature and moisture availability, using Ensemble Empirical Mode Decomposition (EEMD), an adaptive time-space analysis method to decompose trends from oscillatory modes, and Principal Component Analysis (PCA)with regression techniques. Ultimately, their findings observe complex trends and interactions between vegetation and climate, while their statistical approach guards against spurious vegetation signals.
“While increasing land use is known to majorly impact protected areas, less is known about the impact of human-made climate change. This study is among the few to explore possible climate change impacts on Serengeti National Park, which may be increasing in importance over time. This highlights how climate change should be considered alongside land use in protected ecosystems, and provides an analysis framework for other protected areas worldwide,” says Sonali.
“It feels incredible to use data science for generating new knowledge about our environment, in order to inform a sustainable future. I learned so much from Sonali, who gives holistic, physical interpretation of the data patterns from my analysis. The advances of data science, particularly in terms of interpretability and in decision-making process, can benefit tremendously from physical science,” says Teresa.
About the Team:
Teresa Huang graduated from CDS in 2020 with a MS degree in Data Science. She is currently working on her PhD in Applied Mathematics & Statistics at John Hopkins University. Her research interests lie in representation learning, theory in deep Learning, and interdisciplinary research in data science.
Sonali McDermid is Associate Professor of Environmental Studies at NYU. She holds a PhD from the Earth and Environmental Science Department at Columbia University in Atmospheric Science and Climatology. Her research investigates the role landscapes play in our regional and global climate systems.
To read “Moisture and temperature influences” in its entirety, please visit the paper’s IOPScience’s page.
By Ashley C. McDonald
