The main concern when he began the study, says Kato, was that they simply wouldn’t have a long enough data record to detect any trends. Small changes occurring over a large area can have a big impact on climate. But the smaller the change, the longer the data record needs to be to make sure that the change is real. At the time Kato and his colleagues began their study, Terra’s sensors had only accumulated four years of cloud and energy balance data, spanning 2000-2004; Aqua, launched later, had only collected data between 2002-2004. Although he suspected four years of data would be a bare minimum—and probably not enough—Kato figured if the change in the Arctic was dramatic enough, they just might be able to find it.
With those doubts lingering in the back of their minds, Kato and his colleagues began processing, mapping, and analyzing the satellite data. Snow and ice extent data came from a series of microwave-frequency remote sensors nicknamed SSM/I, flown on a series of Department of Defense meteorological satellites. Cloud data came from NASA’s MODIS, and energy budget data came from CERES.
The goal was to create, in essence, a virtual map with three layers that could be laid on top of one another: sea ice and snow, clouds, and albedo. The scientists divided up the Arctic and Antarctic (60-90 degrees North and South, respectively) into a grid in which each box was a bit smaller than 100 kilometers by 100 kilometers. Team members wrote computer software that sorted the satellite observations—usually multiple observations each day— into their appropriate grid box and merged them. Computers continued churning through the data, turning daily observations into monthly averages for the four-year period. Once the virtual maps were complete, the team could test whether changes in the sea ice and snow layer of the virtual map were matched by changes in the maps of clouds or albedo.