MOVING GEOINFORMATICS WORKLOADS INTO THE CLOUD: REAL-WORLD EXAMPLES OF ENABLING COLLABORATIVE RESEARCH

Cloud computing resources have many advantages for geoinformatics researchers: as many collaborative projects have participants who are widely dispersed, hosting data in the cloud and running analytical workloads on virtual machines makes it possible for all participants to work together equally. Equally, the ability to scale computing environments quickly and easily means that small research groups can conduct extremely complex analyses and modelling on very large platforms as needed and then terminate the instances (and associated costs) when the workloads are complete.

We will present two recent examples of geoinformatics projects running on cloud resources.

Seismic surveys for oil and gas exploration generate very large datasets, the analysis of which is very time and resource intensive. A team at the Universidade Federal do Rio Grande do Norte (UFRN) in Brazil, working in collaboration with Oracle for Research, is developing new techniques to improve performance of seismic waveform modelling on these large datasets. To accelerate data processing, we are distributing analytical workloads between CPU and GPU resources. Using cloud computing resources has allowed us to deploy workloads on widely varying combinations of CPU, GPU, and memory to define the optimal distribution of computational work across these heterogeneous clusters. This has led to substantial improvements in the speed and efficiency of the seismic wave modelling, which in turn will reduce the time required to construct and constrain subsurface reservoir models.

We have also partnered with the University of Rajarata in Sri Lanka to model the effects of anthropogenic climate change on marine ecosystems. This work is an important input to policy makers in the region as fisheries and aquaculture are vitally important to much of the local economy. We are combining historic yield data from the Department of Fisheries and Aquatic Resources in Sri Lanka with sea surface temperature (SST) and other data from NASA’s MODIS satellite program. Storing the large datasets in the cloud and running analyses on virtual machines in a data center in Singapore has allowed this important work to continue during local power outages in Sri Lanka and for researchers in the US and Italy to collaborate with them on statistical and GIS analyses of these historic data sets.