Components

Federated Database System (at University of Southern California)

Data in earthquake science

Include observations, simulations, hypotheses
Are geographically distributed and administered independently according to different policies
As the amount of data increases, it becomes a tedious and error-prone task to manually integrate these databases

Toward this end, we suggest a new approach for the Federated Database Services using Ontologies that will provide the base for the semantic wrapping of information sources.

The focus of Federated Database Services is to manage the metadata and incorporate ontologies into the data integration processes.

The approaches to achieve the Federated Database Services are to abstract semantics into an ontology structure and manage federated data with a database structure.

Coupling Methodologies for Cross-Scale Tools (at Jet Propulsion Laboratory and Indiana University)

The approach is to:

Set limits:

applications talking on separate machines
consider real geophysics workflow

Survey cross-scale tools and data sources
Chart communication interactions
Finalize requirements for Complexity Computational Environment (CCE)

Elements include:

Database archives: pulled by code on demand, high latency ok, support OpenGIS.
Sensor Data: pushed streaming data, inline filter, interruptible.
Models/Services: both push and pull, supports range of needs as jobs interact.
Workflow: scripting that accounts for speed/latency/availability needs, including call backs, streaming data, file-based flow.

Complexity Computational Environement and Web Services (at Indiana University)

Elements of the CCE:

Portals and Portlets

We develop component based portal systems as clients

Grid Web Services

We develop both “execution” grid and data grid web services following the WS-I+ model (UK e-Science program).

Messaging and Events

We develop NaradaBrokering for messaging, events, and data streams.

Focus of this work:

Building data grid web services on OGC standards

Support (catalog-based) data needs of AIST applications.
Develop better portlet interfaces for improved human interaction with both data sources and results (overlay maps).

Build NaradaBrokering and Web Service administration and scripting tools.

Script remote AIST application and data service interactions through events.
Support streaming data between services over NaradaBrokering connections.

Data Assimilation Infrastructure (at University of California, Davis)

Data Assimilation Using Evolutionary Grid Search Optimization: Towards Model Steering

Ensemble forecasting methods are only accurate to the degree that observed data has been assimilated into them, allowing for model updating and model steering.

Basic Method: Use parallel grid search, both for model steering and data assimilation.
System (and model) follows a dynamical path through state space, so observations can be used to periodically adjust model parameters, to bring the model path as close as possible to the path represented by the observed system.
Cost function, or fitness function defines the misfit between the model path through time and the path represented by the observed data. Must account both for variation in model parameters, as well as variations in initial conditions for process.
Evolutionary programming approach. Model tuning occurs by parallel grid search, combined with stochastic variation.
Prototype the process using simpler, but still nonlinear, models.

Data Mining Infrastructure (at Jet Propulsion Laboratory)

GPS Time Series Segmentation

Complex data with subtle signals is difficult for humans to analyze, leading to gaps in analysis
HMM segmentation provides an automatic way to focus attention on the most interesting parts of the time series

SCIGN GPS Network Analysis

Have found a way to detect regional aseismic signals
This software is being integrated with the Quakesim web portal
Scenarios for use with real time streaming data through the web portal are currently being investigated

Webmaster: Maggi Glasscoe