Virtually Exploring the U.S.

As a student, your biggest fear going into an internship is that you are going to be a “gopher” — instead of working on projects you spend your time fetching coffee and making copies. However, as an intern for the Center for Conservation Innovation (CCI), it was clear on day one that I was not to be their gopher, but a valued team member tasked with my own work and responsibilities. There was no shortage of work and most days were spent virtually traveling across the continental United States collecting satellite imagery for a new tool that’s in development!

The change detection tool will automatically detect habitat loss using two algorithms that automatically find the changes between two sets of images and it can alert a user as soon as change occurs. This incredible tool is being developed by the CCI team and SkyTruth, a nonprofit technology company, to identify habitat change caused by human disturbances, and it will eventually be available to conservationists around the world. Before its launch, both algorithms must be tested for accuracy and efficiency, and that’s where I came in! Using Google Earth, my job was to find locations all around the United States where habitat loss has occurred due to different kinds of disturbances. It sounds fairly easy, but things got quite challenging when it came to finding the perfect location.

What makes a perfect location? First, it must be a site where human development has replaced natural habitat. To find these places, I wasn’t just randomly clicking on a map. I would start by picking a state and researching it to learn the major habitat types, the disturbances they face, and the best areas to search for them. I would learn all I could about any economic development that would be affecting habitats, filling pages in my notebook with this information.

But just finding a site with development wasn’t enough — for a site to be useful those changes had to happen within a specific time frame. Both change detection algorithms compare a “before” and “after” image, and we’re using satellite images that only go back to 2015. So, for a site to work, the habitat change had to occur after 2015. There were many locations that I passed up, because the alteration happened much earlier. There were also times when a change happened more recently than 2015, but there were not enough quality satellite images between the “before” and “after” dates for the algorithms to show change.

It was really difficult to find that perfect site, and because some areas in the United States don’t have current satellite imagery available, I had to ignore them. Spending hours panning across a state in search of that perfect location was daunting, but the reward was well worth it. Once I found a great site I was hit with such a rush — like finding that one item you’ve been grinding for, for hours in a videogame. Also, I knew once I found a suitable location, chances were there would be two or three more possible sites close by. However, the best part of the process was picking another state with a different type of habitat to search in and starting the quest all over again.

Example of habitat loss due to logging in Cove, OR identified in Google Earth. This is a site at which I was able to test the ability of the automated change detection algorithms to identify this disturbance.

This project became some much more than just an assignment at my internship. I wanted to leave my mark on the Center for Conservation Innovation. I wanted to make them proud. They gave me the opportunity to be much more than a gopher and I wanted to show them that opportunity was not wasted. This has been one of the best internship experiences I have had in college and I know I left my mark. By working hard and making the most of my opportunity I was able to have a true impact, and support Defenders’ mission to protect American wildlife and habitats for future generations to come.

Sites across the United States where the ability of automatic change detection algorithms to identify habitat loss were tested.