Handling the Diversity in the Coming Flood of InSAR Data with the InSAR Scientific Computing Environment

Monday, 15 December 2014
Eric Michael Gurrola1, Paul Alan Rosen1, Gian Franco Sacco1, Piyush S. Agram2, Marco Lavalle2 and Howard A Zebker3, (1)NASA Jet Propulsion Laboratory, Pasadena, CA, United States, (2)Jet Propulsion Laboratory, Pasadena, CA, United States, (3)Stanford University, Stanford, CA, United States
The NASA ESTO-developed InSAR Scientific Computing Environment (ISCE) provides a
computing framework for geodetic image processing for InSAR sensors that is
modular, flexible, and extensible, enabling scientists to reduce measurements
directly from a diverse array of radar satellites and aircraft to new
geophysical products. ISCE can serve as the core of a centralized processing
center to bring Level-0 raw radar data up to Level-3 data products, but is
adaptable to alternative processing approaches for science users interested in
new and different ways to exploit mission data. This is accomplished through
rigorous componentization of processing codes, abstraction and generalization of
data models, and a xml-based input interface with multi-level prioritized
control of the component configurations depending on the science processing
context. The proposed NASA-ISRO SAR (NISAR) Mission would deliver data of
unprecedented quantity and quality, making possible global-scale studies in
climate research, natural hazards, and Earth's ecosystems. ISCE is planned to
become a key element in processing projected NISAR data into higher level data
products, enabling a new class of analyses that take greater advantage of the
long time and large spatial scales of these new data than current approaches.
NISAR would be but one mission in a constellation of radar satellites in the
future delivering such data. ISCE has been incorporated into two prototype
cloud-based systems that have demonstrated its elasticity to addressing larger
data processing problems in a "production" context and its ability to be
controlled by individual science users on the cloud for large data problems.