New Learn ArcGIS Lesson Features Ecological Marine Units

02-27-2018 12:47 PM
Labels (1)
Esri Regular Contributor
0 0 377

The  Learn ArcGIS team is pleased to announce the release of a new Learn lesson featuring Ecological Marine Units data:

Predict Seagrass Habitats with Machine Learning

Seagrasses are an important ally in combatting global warming—these coastal marine plants sequester vast amounts of carbon dioxide. When compared to terrestrial tropical forests, they can store up to 100 times more CO2 per acre. In addition, seagrasses have a large economic value: they provide shelter for marine life such as invertebrates, fish and sea turtles, making them important for local fishing economies. The roots help anchor sediment to the seafloor, decreasing the impact of storms and stopping erosion from affecting coastal homes and businesses. Understanding the habitat of this species is important to drive conservation efforts and map out areas where they disappear due to human interaction.

You are a marine ecologist who wants to model suitable locations for seagrass habitats around the world. Though you only have seagrass data for a small region of Florida, luckily seagrasses tend to grow in similar ocean conditions in coastal areas around the world. Using the predictive powers of a machine learning model along with the spatial analysis capabilities of ArcGIS Pro, you’ll find suitable locations for seagrass growth globally. First, you’ll create a training dataset with all the ocean variables that influence seagrass growth. Then, you’ll put the variables into Python and use a random forest prediction model to determine where the ocean conditions support seagrass growth. Finally, you’ll save the prediction results as a feature class and import it into ArcGIS Pro to find where the highest density of growth is likely to occur.

Skills: Manipulating and cleaning data, Loading machine learning libraries into ArcGIS Pro, Enriching data to fill missing values, Transferring data between ArcGIS Pro and Python, Performing analysis in Python, Performing prediction using random forests, Using geoprocessing tools for statistical analysis.

Thanks to Orhun Aydin‌, Marjean Pobuda‌, Keith VanGraafeiland, and Kathy Cappelli for their excellent work in developing this lesson.

About the Author
Dawn was appointed Chief Scientist of Esri in October 2011 after 17 years as a professor of geography and oceanography at Oregon State University. As Esri Chief Scientist, she reports directly to Esri CEO Jack Dangermond with a mission to strengthen the scientific foundation for Esri software and services, while representing Esri to the national and international scientific community. Dawn maintains an affiliated faculty appointment as Professor of Geography and Oceanography at Oregon State. Follow her on Twitter @deepseadawn. More info. also at and