Amazon Web Services has partnered with North Carolina State University’s Institute for Climate Studies to develop cloud-based datasets for two studies on the Earth's natural phenomena using tools such as the Amazon Elastic Compute Cloud.
The first project was part of an international effort to assess the Earth’s ability to reflect solar radiation or albedo, while the second involved the use of recorded temperature data to create a historical climate dataset for the National Oceanic and Atmospheric Administration, AWS said Sunday.
As part of the first study, the scientists tested EC2’s capacity for developing an albedo dataset from over 270 terabytes of NOAA geostationary satellite data and found that AWS can complete the task 50 times faster compared to an on-premises system.
For the second study, called “Global Historical Climatology Network–Monthly”, scientists had to process thousands of observation data from weather stations around the world and make adjustments based on factors such as instrumentation, location and observation methods. The entire effort took around 6.5 days instead of one month as expected with local systems.
The team spent around $13K for 20 hours of AWS computing time for the first project and around $142 for the second effort.