• Skip to Content
  • Skip to Main Navigation
  • Skip to Local Navigation
  • Skip to Search
  • Skip to Sitemap
  • Skip to Footer
Help develop a Bird ID tool!
Around the Lab

Multivariate Madness

article spread
by Pat Leonard

So many forces are at work in an ecosystem, it is difficult to study or understand how they all interact—often called “multivariate madness.” But a new partnership between ornithologists and computer scientists is bringing us a step closer to understanding the variables that influence bird distributions and migration across North America.

The Cornell Lab is combining more than 48 million bird observations gathered in eBird with information about weather, habitat, and plant greening collected by remote-sensing NASA satellites.

“Our goal is to better understand how habitat and other environmental cues influence the patterns of bird migration,” says Steve Kelling, director of Information Science at the Cornell Lab. “We hope to forecast how bird migration patterns might change as the environment changes across North America.” Kelling is one of the key players in the National Science Foundation’s DataONE project and a member of Cornell’s Institute of Computational Sustainability, both of which are developing new techniques for combining and visualizing large sets of environmental data.

It takes a powerful computing infrastructure to synthesize and analyze this much information and translate it into visual models, but the massive supercomputers of the TeraGrid are up to the task. “TeraGrid is a network of supercomputers run by the National Science Foundation,” says Kelling. “We could have 1,400 processors of a TeraGrid supercomputer working simultaneously to process all our data for one species and have it done in a few days.” Kelling hopes to do this for all 700+ species of birds in North America, and 100,000 hours of processing time on the TeraGrid have been allocated for the work.

“One of the exciting things we can do with the models we create is to run computer simulations that shift when environmental patterns occur, such as when leaves become green in the spring. Then we can estimate how a species would need to respond in various climate change scenarios,” says Kelling. “By bringing high-performance computing techniques into the realm of conservation biology, we can begin to tease apart the most important variables influencing when and where bird species occur. We’re discovering patterns in existing data that were hidden before we found a way to analyze this huge amount of information.”

—Pat Leonard

Living Bird Magazine

Autumn 2010

Table of Contents