Technical Significance
NASA sensors collect TB of data per day that cannot be manually processed and analyzed. New architectures and algorithms are needed to facilitate rapid, autonomous analysis and utilization of sensor web data streams. This research is:

• Demonstrating an automated capability for discovery of causal relationships in a large, distributed, mixed format database.
• Developing a proven architecture for automated retrieval, processing, analysis, and rapid utilization of distributed, heterogeneous, remote sensing data.
• Demonstrating an automated planning capability which generates optimized data selection, processing, and analysis plans based on user specified criteria and goals.
• Generating real-time, regional to global scale maps of biospheric parameters and forecasts of episodic biospheric events (e.g, fire risk maps) using a system which incorporates autonomously discovered causal models.
• Incorporating machine learning techniques into the integrated system to provide continuous learning and model correction functionality.