(abstract for IGARSS 2002)

MODster: Peer-to-Peer Sharing for Remote Sensing Standard Products

JAMES FREW
Donald Bren School of Environmental Science and Management
University of California, Santa Barbara

Standard data products from sensors on NASA's Earth Observing System
(EOS) satellites (Terra and Aqua) are currently available only from a
few EOS data centers. These centers can become distribution bottlenecks
if demand for the products increases, or if the centers' data systems
develop problems. Moreover, a few data centers cannot satisfy myriad
demands for product customization (subsetting, re-formatting,
re-projecting, etc.) Since EOS data are not copyrighted, there is no
reason why non-EOS data centers, or even individual scientists, cannot
redistribute whatever EOS data granules (basic units of distribution)
they possess, and/or offer value-added transformations of those
granules.

In much the same way that the popular Napster service allowed
participants to share well-known music files, we are building MODster to
allow participants to share well-known data granules from the EOS
Moderate Resolution Imaging Spectrometer (MODIS). By "well-known" we
mean that the item to be shared adheres to a common naming scheme:
artist and title for music; EOS "local granule ID" for MODIS. Thus a
particular local granule ID always refers to the same data content,
regardless of which sites actually house copies of the granule.

A central Web server provides an up-to-date catalog of all MODIS
products made available by any and all participating sites. The only
requirement for these participants is that they make their MODIS
granules available via HTTP, at URLs that they provide to the MODster
server. The server's configuration database maintains [product,
granule, URL] associations and uses them to redirect MODIS granule
requests to whichever participating site can satisfy them.

Participants in MODster thus have a standard way to publish their MODIS
data holdings to the broader community, and may in turn access the
entire community's holdings through a single "namespace" (the MODSter
base URL plus the MODIS local granule ID).

Direct delivery of data granules via HTTP lends itself easily to the
superposition of value-added services. We are experimenting with two of
these services in particular: the Distributed Ocean Data System (DODS)
and the Web Mapping Services (WMS), which provide standard
specifications for common operations like subsetting, and which can be
directly connected to data analysis environments like MATLAB and
ArcView.