Title: Clear and Precise Definitions of Scientific Processes to Facilitate the Conduct of Science

Speaker: Dr Rodion Podorozhny

Time: 12:30pm-1:30pm, April 4th, 2008

Location: Nueces Conference Room

Abstract:

Sensor networks that provide high resolution spatial and temporal measurements 
will soon support real-time environmental modeling and forecasting.  
But the reliability of the resulting datasets (including both original 
measurements and derivative models and forecasts) will depend on the ability 
to reproduce the processes used to create them and to verify that these 
processes are scientifically sound.  

We are developing cyberinfrastructure tools that support precise description 
and execution of processes, based on a formal process definition called an 
"analytic web."  This approach guarantees dataset reproducibility by 
providing (1) detailed process metadata that precisely describes all 
sub-processes, (2) a complete audit trail of all artifacts (e.g., datasets, 
code, models) used or created in a particular execution of a process, and 
(3) annotation of these artifacts with the appropriate process metadata.  
It also supports rigorous evaluation of processes for errors, including 
logical, statistical, and propagation of measurement errors.  These tools 
are being incorporated into a sensor network that will integrate 
meteorological, hydrological, eddy flux, and tree physiological measurements 
to study the movement of water through a forest ecosystem.  The system is 
designed to provide optimal real-time data and process metadata for modeling 
and forecasting.  

Features will include: (1) real-time quality control, modeling, and 
gap filling; (2) scheduled post-processing to update models with subsequent 
measurements; and (3) facilities for substituting corrected or alternate 
measurements as needed.  The analytic web tools are specifically designed 
to handle the complex process features of such a system, including 
concurrency, real-time data streaming, and exception handling.  These are 
features that we believe will be commonplace in sensor networks of the future.