Mars Data Mining
by: Rod Rice, August 14, 2006 5:08:00 am
Among the data being sent back from Mars are many images of the red planet's surface. In the past the way that information was analyzed was by a process called geomorphology which is essentially describing what is seen in the pictures. Because this is a very slow process only a fraction of the available data has been analyzed so far. Enter University of Houston computer scientist Ricardo Vilalta. He is using computers to classify various landscape features.
"The way we do that is, we ask the expert to label, manually, some of the landscape formations on Mars, and then we look for patterns or correlations among the features characterizing these landscapes. After that we let the computer classify new landscapes that we haven't seen before automatically."
Professor Vilalta is working with Tomasz Stepinski of the Lunar and Planetary Institute in Clear Lake. He is interested in exploring Mars past. It is now a cold dry planet but once it may have been warmer and wetter meaning it may have been able to sustain life. He's looking for evidence that indicates erosion on Mars may be similar to erosion on earth caused by river systems. Dr Stepinski says that would be a place for further exploration.
"The decision for NASA would be to send a spacecraft to a specific location when they would have the best chance to study whether or not there is any evidence of past life."
Professor Vilalta says data mining has a lot of applications in business and industry. In this process though scientists are using computers to get a lot a data about a small section of Mars to learn more about a larger area of Mars.
"And the way we do that is by using a cluster of computers, about 150 computers all connected together, and we try to do that in parallel so that all the processing can be done in a relatively small amount of time. Instead of weeks we're talking about hours."
So far the data mining has yielded information about what is believed to be one of the oldest surfaces of Mars that is a bit surprising for planetary scientist Stepinski.
"At least in the area of Mars that we are looking at there is not much variability. The next place looks pretty much the same as the one we initially started. So it's pretty uniform."
Unlike Earth where there are mountains, valleys and plains.
This work is currently being funded by a quarter million dollar grant from NASA's Applied Information Systems Department.