Zooming from her office, MSU researcher Kelly Kapsar showed off stunning satellite images from across the United States.
In some, different colors revealed the precise heights of the forested ridges in the Sierra Nevada or the craggy peaks of the Rockies and the valleys between them. Others showed the pine flatwoods of southern Florida or the smooth rolling grasslands of the Kansas prairie.
The images were taken by the space shuttle Endeavor during a 2000 mission to make detailed radar maps of the planet’s surface.
Endeavour orbited Earth 16 times a day during the 11-day mission. In that time, it took more than a trillion elevation measurements, generating a whopping 12.3 terabytes of data.
Since then, other Earth-observing satellites have measured things like temperature, rainfall, sea levels, carbon dioxide, snow cover, wind speeds, even dust — and at scales ranging from the span of a continent to patches barely larger than a tennis court.
It’s a treasure trove of data. And by combining this information captured from space with measurements of plants and animals taken on the ground, researchers hope to better predict where species are most likely to thrive in the years to come, said professor Phoebe Zarnetske, who directs MSU’s Spatial and Community Ecology, or SpaCE, Lab.
But there’s a hitch, said Kapsar, a postdoctoral associate in the SpaCE Lab.
While a lot of satellite remote sensing data is publicly available, for many researchers, taking advantage of it isn’t straightforward.
“We have so much satellite-based data now, but ecologists receive very little training on how to work with it,” Kapsar said. “Once you start downloading satellite data, it gets into hundreds of gigabytes and thousands of layers.”
That often requires technical expertise in supercomputing and big data analysis to process.
There’s another problem, Kapsar added. The statistical models that scientists use to map where plants and animals are likely to live often require a single, numerical value to characterize the environment. But from the perspective of, say, a vole or a beetle, summarizing a landscape in just one number doesn’t tell the whole story.
To illustrate, Kapsar pulled up a series of satellite images showing the contours of the land around sites within the National Ecological Observatory Network, or NEON, a 30-year research effort to monitor changes at 81 field sites across the U.S.
One NEON field site within the Sierra Nevada mountain range in California has an average elevation of 7,050 feet. However, this number masks a lot of ups and downs, from towering 10,000-foot mountain peaks to valleys and meadows.
The same holds true when it comes to rainfall, Kapsar said. While annual precipitation in a NEON site near Las Cruces, New Mexico, in the northern Chihuahuan Desert stays pretty consistent around 11 inches a year, rainfall at a site in Hawaii can vary drastically year to year — from 80 inches to 160 inches.
Also known as “geodiversity,” this heterogeneity in the environment can mean a lot to the plants and animals that live there.
“If you think about the world like many animals do, they’re not going to just take the average of a mountain landscape,” Kapsar said. “They have to contend with things like: How steep is this cliff? Can I climb it? How many ups and downs are there? How bumpy is the terrain? Are there places for me to hide?”
“The temperature and precipitation within an area can vary with the topography and vegetation to produce microclimates, where only certain species can thrive,” said co-author Lala Kounta, a climate scientist in the SpaCE Lab.
Kapsar, Kounta, Zarnetske and collaborators are working on ways to provide geodiversity data to researchers and help close the gap.
Crunching massive amounts of satellite data with the MSU High Performance Computing Center in a new study, they hand over a set of climate and elevation geodiversity metrics that capture more than just the mean so researchers working at NEON sites across the U.S. can use them.
Originally developed for the field of surface metrology, the metrics include statistical measures of how “rough” or “smooth” precipitation, temperature and elevation are across the landscape, calculated using an open-source computer program that Zarnetske and MSU’s Kyla Dahlin and colleagues developed called GEODIV.
Their paper also offers a how-to for researchers who want to use satellite data to capture this complexity at other sites around the world and at different scales.
“The idea is to take the satellite’s perspective from way up high in the sky, and NEON’s intensive data collection on the ground — catching bugs, listening for birds, sampling plants — and bring them together to get the best of both worlds,” Kapsar said.
