Machine Learning Based Seagrass Growth Prediction for Mitigation and Habitat Assessment

Coastal development and infrastructure projects often require seagrass mitigation and impact assessments. Seagrass growth depends on many factors, including sunlight penetration into the water column, water chemistry, hydrodynamic forcing, and other factors which can make assessing impacts to existing seagrass beds or designing a successful mitigation project difficult.  This paper proposes a methodology for implementing a machine learning model to efficiently predict potential seagrass growth areas in inland Texas Bays. The machine learning algorithms can be used to quickly assess mitigation site viability and project performance, or to assess impacts to existing seagrass beds. Two separate approaches were taken—developing large scale and site-specific machine learning models.  The large-scale machine learning model can be used during mitigation site assessment to select areas with high potential for seagrass growth. This can be used to select a site for a mitigation project where seagrass growth is likely if a proper mitigation project is designed and implemented. The small-scale tool can then be used to evaluate the different mitigation project alternatives at the selected site to optimize project design and maximize the likelihood of seagrass growth. The final machine learning algorithms found between 73% and 95% spatial accuracy in predicting the growth of seagrass. In combination the large scale and small-scale models provide a database for rapid assessment of seagrass growth.