KML Communicates CO2 Findings

March 20, 2009 By: Cyrena Respini-Irwin

A Google competition showcases the scientific applications of Keyhole Markup Language.

Earlier this month, Google announced the winners of its KML in Research contest, which challenged entrants to apply Keyhole Markup Language (KML) to scientific research. (As many Geospatial Solutions readers will recall, KML is a programming language that can display geographic features in Google Earth and other applications.) The contest's seven winners included both students and professionals, and their projects ranged from visualizing precipitation forecasts to distributing LIDAR models.

One of those projects focused on carbon dioxide levels — a topic that should interest anyone who's read the news lately. "North American Carbon" was the winning entry from Tyler Erickson, a research scientist at the Michigan Tech Research Institute. He's also an adjunct assistant professor in Michigan Technological University's Department of Civil and Environmental Engineering.

Erickson modeled CO2 uptake and release cycles in KML, working from massive datasets created by researchers at the University of Michigan and the Global Monitoring Division of the National Oceanic and Atmospheric Administration's (NOAA) Earth System Research Laboratory. NOAA collects CO2 measurements with instruments mounted on cell phone transmission towers and similar structures. That data is then employed in Assistant Professor Anna Michalak's research group at the University of Michigan, where the goal is to predict future changes in carbon "sinks" — such as forests and oceans — by determining where CO2 is being absorbed from the air or released into it.

This kind of research generates large volumes of data that vary in location, as well as time. By formatting that data with KML, Erickson enabled researchers, decision makers, and the general public to explore it in the Google Earth interface. "KML is very useful for communicating data to a wide audience," he said, noting the language's widespread use and the public's familiarity with Google Earth. Erickson added that the capability to include annotations, such as video clips, can introduce non-technical users to the underlying science involved.

The Virtue of Sharing

Erickson specializes in building geospatial systems with data that varies spatiotemporally, but some of those data are more restricted than others. "It was interesting to apply these tools to a dataset that's more in the public interest," he said, noting that unlike some U.S. Department of Agriculture and Department of Transportation projects he's worked on, the measured CO2 data is accessible to everyone. "It's from a community that's very used to openly sharing their data," Erickson observed. Atmospheric research, he explained, is a subject that by its very nature necessitates collaboration across geographic regions and political boundaries.

Erickson is a proponent of data sharing in science; in fact, he would like it to be a requirement for researchers who receive government funding: "I'd like to see [data sharing] as the default, and restrict it if there's a reason to, instead of the other way around." And his collaborative tendencies don't end there — they extend to tools as well. "It's very nice when you can apply [open-source geospatial software] to open data that can be shared in an open manner."

Although Erickson uses a combination of proprietary and open-source tools in his work, he is migrating toward more of the latter. In addition to placing a greater emphasis on interoperability, he explained, open-source offerings can be altered by a knowledgeable user to get around the inevitable roadblocks that can stall a project. "If you use open-source tools, there's always a way forward."

More Data, More Time

Despite his project's blue-ribbon status, Erickson considers it just an initial step in using KML for carbon science; he hopes to garner the funding to expand upon this prototype, making it more valuable to the research community. For example, instead of being limited to one month's worth of data, the model could incorporate the complete time series of measurements and model results instead. Other goals include adding more data layers, allowing researchers to do more exploratory analysis of their data, and providing more visualization of the uncertainty in the data.

Bookmark it: digg del.icio.us technorati yahoo facebook twitter

Add Comment