Skip to content

PREDICTING THE IMPACT OF RIPARIAN VEGETATION AND LAND USE ON STREAM TEMPERATURE IN THE CHESAPEAKE BAY WATERSHED USING DEEP LEARNING

    Mahmod Por, Elham, Department of Civil and Environmental Engineering, The Pennsylvania State University, University Park, PA 16802,ebm5774@psu.edu; Van Meter, Kimberly, Department of Geography, The Pennsylvania State University, University Park, PA 16802, vanmeterkvm@psu.edu.

    Stream temperature is a fundamental driver of river ecosystem health, with profound implications for biodiversity, water quality, and the resilience of aquatic habitats. Understanding the intricate interplay between environmental factors—such as riparian vegetation, land use, and climatic forces—has never been more critical, as urbanization and climate change pose escalating threats to watershed integrity. This study presents a novel application of Long Short-Term Memory (LSTM) deep-learning models to investigate how both local riparian buffers and upstream land use patterns influence stream temperature across the expansive Chesapeake Bay Watershed (CBW). Harnessing high-resolution (1-m) land cover data in conjunction with 30-m National Land Cover Database inputs, we demonstrate the power of deep learning to capture spatial and temporal complexities in stream temperature regulation. Our results reveal key stream reaches where riparian deforestation and land-use changes intensify thermal stress, contributing to ecosystem degradation. By identifying these critical areas, our study not only enhances our understanding of watershed-scale environmental processes but also provides actionable insights for conservation and restoration priorities. This research underscores the role of advanced modeling techniques in shaping sustainable watershed management strategies, offering a scientific foundation for adaptive responses to ongoing climate and land-use pressures. The implications extend beyond ecological preservation, providing a framework to safeguard both river health and the well-being of communities reliant on the Chesapeake Bay and its tributaries. Our findings serve as a call to action for integrated, science-driven management aimed at fostering resilient ecosystems and sustainable river networks in the face of an uncertain environmental future.

    Stream Temperature, Chesapeake Bay Watershed, Deep Learning, Climate Change