|  1.
Modeling and Computational Priorities
The broad range of spatial and temporal scales manifested by solid-Earth
processes calls for a variety of modeling and data assimilation
techniques. Advances in inversionmethods, three-dimensional modeling,
data assimilation, statistical analysis, and pattern recognition
are all necessary for understanding these complex systems. High-performance
computers are required for carrying out these approaches.
One
of the major problems facing scientists today is that the scientific
data volumes are increasing at a faster rate than computational
power, challenging both the analysis and the modeling of observations.
Resources must be put into improved algorithms to simplify processing
and to approximate complex phenomena to allow researchers to handle
the large volumes of data as well as to find the dominant physics
in a given data set. Another promising approach to handling large
data volumes is to use pattern recognition to focus attention and
point out subtle features in the data.
Because
of the complexity of the solid-Earth system, high-performance computers
are required for scientific progress. Computations of the systems
being studied, from the geodynamo to interacting fault systems,
take weeks to years to run on even the most capable of current workstations,
making supercomputers the only means of modeling the systems. It
is crucial to utilize the latest computational advances to make
modeling an effective tool.
2.
Distributed Receiving and Processing Systems
An important aspect of data collection is to create distributed
centers for processing and storing unique data sets. Developing
the infrastructure to compare and use complementary data sets, such
as ice topography and sea-level changes, opens the door to interdisciplinary
research. It is also important to create the infrastructure to access
other non-NASA datasets such as seismic and geologic data. These
supporting data sets are critical for modeling and understanding
the complete system. These distributed data centers are particularly
important in the event of natural disasters, when not only can they
support disaster management but they also enable real-time scientific
experiments dependent on time-sensitive observations. Such centers
will become more important as multiple data types are fused into
integrated models. |
