RII Track-4:NSF: The Monitoring of Invasive Yellow Sweet Clover Using Landsat/Sentinel-2, UAV Imagery and Machine Learning

Invasive yellow sweet clover (YSC) is an annual legume herbaceous flowering plant, other than grass, planted initially for bee habitat and soil erosion management. YSC can also cause hemorrhaging and poisoning in livestock as a hay component. The goal of this fellowship is to initiate a long-term collaboration between the University of South Dakota (USD, home) and the United States Geological Survey (USGS) Earth Resources Observation and Science Center (EROS, host) for the mapping of YSC blooms.

The PI and graduate student, with the help of Host scientists, will jointly develop machine learning predictive models using High-Performance Computing (HPC). The knowledge transfer will involve the adaptation of the EROS processing chain to the supercomputer at USD, which would immediately improve computational ability and the long-term competitiveness of USD in invasive plant species mapping.

The research methods developed here will enable the production of species distribution maps to inform land managers and policymakers to help manage the rapid spread of YSC across South Dakota (SD) and the Northern Great Plains. The proposed product would have long-term impacts on STEM education at USD by providing a series of topics for undergraduate research, master’s theses, and Ph.D. dissertations.

Students could also collaborate with USGS scientists at a world-class Federal Lab, leading to summer internships and possibly full-time job offers.

This Research Infrastructure Improvement Track-4 EPSCoR Research Fellows (RII Track-4) proposal would provide a fellowship to an Assistant Professor and training for a graduate student at the University of South Dakota. This work would be conducted in collaboration with researchers at the USGS Earth Resources Observation and Science Center. There has been a dramatic increase of Yellow Sweet Clover (Melilotus officinalis; YSC) with super blooms in SD and the Northern Great Plains (NGP) following higher precipitation in recent years.

YSC has the potential for establishing significant biomass in its biennial lifecycle and provides competition to native grass species through shading. There are major knowledge and data gaps regarding the drivers, spatiotemporal extent, or tipping points of YSC blooms. Hence, near-real-time mapping tools, at a broad spatial scale and high resolution would be helpful in identifying drivers and enabling targeted monitoring and management of YSC.

Our objectives are 1) to develop an annual YSC % cover predictive model for SD and train on field samples of YSC along with site-specific variables (topography, land cover, soil moisture, and edaphic factors) and climate to optimize model estimates; 2) The model parameters will be applied to the Sentinel-2 (HLS) time series to produce a time series of annual YSC % cover and biomass maps. The PI will adapt and evaluate USGS EROS process flows to develop high-resolution vegetation mapping capabilities that include three future scenarios: wetter & cooler, normal, hotter and drier, from short, mid, and long-term weather forecasts.

These maps could be used to detect year-to-year changes in YSC and enable species distribution maps to inform land managers and policymakers to help manage the rapid spread of YSC across SD and the NGP. Methods developed could serve as prototypes to map other invasive plant species as well as structure and function attributes of rangeland vegetation leading to new opportunities and innovations.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.