LarvaMap— particles in the (compute) cloud

LarvaMap is an end-user application to run particle tracking models based upon the ROMS ocean circulation model. LarvaMap uses data stored in a cloud-based data store and the application is installed in the cloud. While the application specifically models the transport of larval fish, it currently treats them as inanimate particles carried on ocean currents. The application could similarly be used to model the transport of dust, contaminant plumes, and other atmospheric phenomena.

The front end of LarvaMap uses an Adobe Flex thin client with the OpenScales mapping Application Programming Interface (API), to provide an exploratory map interface, and access to the Environmental Data Server (EDS) and metocean Web Map Services (WMS) for contextual wind and current data. The user defines larval release parameters, including temporal extent, release duration, number of particles, and species. Utilizing a Java-based distributed data retrieval and model processing backend, the larval transport model runs as a service, dynamically accessing oceanographic data from distributed THREDDS servers.

The project is funded by NOAA's High Performance Computing and Communications office (HPCC) and evaluates how distributed services-based systems can be used to improve computationally intensive science projects. The LarvaMap system has been deployed on the Amazon cloud and may also be deployed on one or more other cloud computing architectures. LarvaMap is a good application to test both the computing aspects of the cloud and the storage and transport of large datasets needed to run models. In addition to deploying the system in the cloud, future developments include additional inputs for defining species and adding in larval fish behaviors, such as vertical migration.

Tools such as LarvaMap can be used to explore implementing existing applications in compute and data cloud resources. They can also be used to communicate the results of modeling by making it possible for the end-user to easily run particle tracking models to test scenarios and to analyze the results of model runs. By making a cloud computing resource easily accessible via a web interface, it opens up the use of these massive processing resources to non-modelers.