Using remote sensing and geospatial analysis to understand changes to Lake Michigan dunes

Remote sensing and geospatial analysis techniques allow for better understanding of dynamic landforms such as sand dunes. Our study investigated to what extent geospatial analysis of historic aerial photographs could be used to detect changes in dune activity over time at three Lake Michigan parabolic dunes in western Michigan, USA. We georeferenced historic aerial photographs, dating from 1938 to 2008, and then used edge-detection in remote sensing software ERDAS Imagine to distinguish bare sand from vegetated areas. The photos were then imported into the geographic information system ArcGIS, where they were reclassified and vectorized to create bare-sand outlines. Further analysis in ArcGIS allowed us to determine the changes over time to the bare-sand areas and the movement of the edges between bare sand and vegetation along the central axis of the dunes. Results show significant variability in each dune's bare-sand area during the study period, although only small increases in bare-sand area were recorded from the beginning to the end of the study. An indicator of continuing dune activity is the eastward advance of the inland edge of each bare-sand area from 1938 to 2008. Understanding changes in dune activity, especially long-term variation, through utilizing these geospatial technologies offers new insights and opportunities for research in the ongoing study and management of west Michigan dunes.