Overview

Datacasting is an RSS-based technology for distributing Earth Science data.  Providers of the data publish the availability of files through a web-feed, along with relevant ancillary information pertinent to Earth Science data (e.g., file format, data collection methods and data content).  Users subscribe to the feeds with the Datacasting Feed Reader, which enables them to list and interrogate the feeds for identification and download of the files for further analysis.   

A powerful addition to the Datacasting Feed Reader is that of filtering, which gives users the ability to precisely identify the files that are relevant to a particular need.  By building filters that make comparisons with information contained within a feed, users are able to construct lists of relevant files and have these files downloaded automatically.     

For example, a user might subscribe to a Datacasting feed that contains information about files produced by an orbiting imaging satellite, but they may only be interested in data that contain wild fires in California.  The user would therefore construct a filter that lists only the files that have been tagged in the Datacasting feed to contain data related to a wild fire and imaged within a bounding box ().    The user could further refine the filter to show only the files where the wild fire exceeds a specific size or within a certain distance of an urban area.

The types of filters a user can build are solely dependent on the richness of information tagged in the web-feed.  Our hope is that through the Datacasting forum, users are able to make recommendations to data providers on the information that ought to be contained within a feed and also promote the uptake of standard metadata conventions and taxonomies, and thereby enable filtering across multiple feeds. 

In this way, a user interested in identifying all satellite and buoy data that are contained within a specific region and during a specific period that contain information about a harmful algae bloom can identify data relating to their interests. They could even extract higher-order information by building filters that provide answers to questions such as: Are there any HABs currently identified in my area of interest?

In addition to filtering, the Datacasting Feed Reader will have the ability to read files that have been downloaded and display the data.  The intent is to give users the capability to quickly analyze the data and further decide if it is useful or not.    For more in depth investigations, users would use their usual tools to perform analyses on the downloaded data.

Datacasting Feeds

Datacasting feeds are specified using XML, which conform to the RSS specification with additional extensions that are relevant to Earth Science data. These Datacasting extensions contain elements (tags) that describe the data collection (e.g., location in time and space, sensor and measurement retrievals), the file formats (based on ESML, this allows for files to be read and displayed in the Datacasting Feed Reader) and custom elements. Custom elements enable data providers to define metadata that are unique to their Datacasting feed. For example, adding a custom element that quantifies the percentage cloud free pixels in a image enables users to quickly select images which have few clouds. A custom element consists of a definition which can either be of a type float, integer, string, Boolean, region or time.

For a complete description of the Datacasting XML feed specification go here.

In order to help data providers easily build Datacasting feeds, we have develop a set of python scripts that can be easily configured.  These scripts can be downloaded from http://datacasting.jpl.nasa.gov/download.

Datacasting Feed Reader

The Datacsting Feed Reader is written in Java and modeled on a lightweight mail tool.    The Feed Reader provides the functionality to subscribe to feeds, build filters and download files.  The App is fully RSS-compliant.

The Datacasting Feed Reader can be downloaded here.
 

Directory of Datacasting Feeds

The Datacasting Directory provides a listing of Datacasting feeds.  Data providers are encourage to upload details of their feeds, including custom elements, in order for users to search for availble feeds.

The Datacasting Directory can be found here.

Datacasting Team

The Datacasting concept and tools have been developed by a small team of scientists and engineers at NASA's Jet Propulsion Laboratory. The participants include:

     Andrew Bingham, Principal Investigator
     Bob Deen, co-Investigator
     Kevin Hussey, co-Investigator
     Tim Stough, Cognizant Engineer
     Sean McCleese, Developer
     Nick Toole, Developer

The team can be contacted at datacasting@list.jpl.nasa.gov.

Acknowledgements

Datacasting is funded through the NASA ACCESS program here.

This website has been cleared by JPL Document Review, clearance number 07-3575.