Supercomputing infrastructure is helping scientists from the University of Technology Sydney produce satellite data products for vegetation and land monitoring. The researchers have analysed terabytes of Australian regional data from the Terra Modis satellite and brought it all together so that seasonal changes in vegetation can be identified much more easily.

Other satellite data products exist around the world, including one from NASA, but they are often inadequate for dealing with Australian conditions. Our highly variable rainfall and climate extremes make Australian data very different to work with than the Northern Hemisphere climates.

Dr Rakhesh Devadas from UTS says that "We wanted to develop something separately for Australia because Australian conditions are completely different, and the algorithms they use for developing the NASA product may not be completely effective in the Australian context. Our own product is more applicable to Australian vegetation."

High Performance Computing resources at NCI and UTS were used to produce the final data product, which can be found here. Running the image processing algorithms on up to 15 terabytes of image data is too much for any standard computer to deal with, but supercomputers have much less trouble. "This is definitely helping us to proceed further, it's great to have a facility like that," says Dr Devadas.

The next step will be to integrate the system with NCI's Research Data Collections. The aim is to simplify the data products so that researchers with less experience in the computational area can still make use of the information. From there, Dr Devadas says, "We can't really tell what will be an application for these things. It depends on how the researcher wants to use it."

The wide range of uses might include mapping cropping cycles and woody growth patterns, tracking grass pollen release for allergy treatment and monitoring vegetation for fire prevention. With a collection of images of the entire country covering 13 years combined into an accessible data product, numerous environmental monitoring questions can now be studied by many more people than before.

This research project is conducted by the Remote Sensing Group at UTS under the leadership of Prof. Alfredo Huete with funding support from TERN-AusCover.