Communications of the ACM
Big Data Is Too Big For Scientists to Handle Alone
Scientists using space-based observations to study global ecosystems have been overwhelmed by the "sheer number of different measurements required to address the key science questions."
Credit: Peter DaSilva/QUANTA Magazine
The U.S. National Ecological Observatory Network (NEON) seeks to understand the impact of global climate change, land use, and biodiversity on natural and managed ecosystems and the biosphere. David Schimel, who designed NEON, initially was overwhelmed by the "sheer number of different measurements required to address the key science questions." He formed "tiger teams" across the country to develop scientific methodologies and data-processing requirements, and began building more than 100 U.S. data-collection sites with the goal of recording 600 billion raw measurements a year for 30 years. Data collection will begin in 2017, and data will be converted into user-friendly data products that will be freely available to scientists and the public.
However, NEON faces an exceptional challenge in making sense of its data, because it has more than 500 quantities to track, ranging from temperature, soil, and water measurements to aerial imaging and remote sensing. In addition, much of the data, such as taxonomic names, are unstructured and hard to parse.
Experts say big data and distributed computing can only be leveraged in future scientific endeavors with a combination of science, statistics, computers, mathematics, and leadership. "Machines are not going to organize data science research," says University of California, Berkeley professor Bin Yu. "Humans have to lead the way."
From Wired News
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found