WVU Today
New research from West Virginia University is transforming understanding of the Mountain State’s famous landscape — and identifying ways to preserve it.
As WVU’s latest NSF CAREER Award winner, assistant professor of geography Aaron Maxwell will use big data to map what the surface of West Virginia looked like over the last 60 years. The funding includes $636,785 over five years.
Maxwell will use data analytics and advanced computational methods to extract valuable information from existing geospatial data over wide regions of the state. Through these methods, his research can increase our understanding of natural landscapes and how they change over time.
“West Virginia is an interesting place to do this type of work because of the rich, complicated history with surface mining and, on a larger scale, with mountaintop mining in the southern region,” Maxwell said. “There have been massive changes in just the lay of the land, flattening mountaintops and filling in valleys. Having older data and new data allows you to look at change to see where there have been gains and losses in elevation.”
Most mapping techniques to date are dominated by manual methods, including illustrations. However, recent advances in imaging technology through remote sensing applications, like those used by Maxwell, have the potential to revolutionize this process.
“This project uses deep learning, which is the process of training a computer algorithm to recognize certain types of features on the landscape by feeding it lots of examples,” Maxwell said. “Once it has learned, you can apply that knowledge to other locations to automate the process so you can expand the available data and the maps that are generated.”
Because mining and other landscape disturbances reshape the land’s physical surface, automated mapping methods for characterizing those features is urgently needed to understand and manage natural hazards, such as landslides and river erosion.
“West Virginia has one of the highest per capita landslide rates, so there is a good landscape to study that hazard,” Maxwell said. “We are doing a landslide inventory and doing predictive modeling of where landslides might occur in the future.”
As part of the project, Maxwell will partner with WVUteach and CodeWV to create data-focused lesson plans for West Virginia high school environmental geoscience and computer science teachers. The curriculum will focus on the coding languages R and Python, commonly used by data scientists.
“We’re going to get them started in coding and just playing around with data science and spatial data sets,” Maxwell said. “We deal with really complicated data, and there are a lot of careers directly in that or related to it. If you can work with this type of data, you could get into working with business analytics or other similar professions.”
TWEET @DominionPostWV
