Proceedings Article | 25 August 2008
KEYWORDS: Remote sensing, Satellites, Algorithm development, Sensing systems, Sensors, Space operations, Computer architecture, Data archive systems, Data modeling, Decision support systems
In modern horror fiction, zombies are generally undead corpses brought back from the dead by supernatural or
scientific means, and are rarely under anyone's direct control. They typically have very limited intelligence, and
hunger for the flesh of the living [1].
Typical spectroradiometric or hyperspectral instruments
providess calibrated radiances for a number of remote
sensing algorithms. The algorithms typically must meet
specified latency and availability requirements while
yielding products at the required quality. These systems,
whether research, operational, or a hybrid, are typically
cost constrained. Complexity of the algorithms can be
high, and may evolve and mature over time as sensor
characterization changes, product validation occurs, and
areas of scientific basis improvement are identified and
completed. This suggests the need for a systems
engineering process for algorithm maintenance that is
agile, cost efficient, repeatable, and predictable.
Experience on remote sensing science data systems
suggests the benefits of "plug-n-play" concepts of
operation. The concept, while intuitively simple, can be
challenging to implement in practice. The use of zombie
algorithms-empty shells that outwardly resemble the
form, fit, and function of a "complete" algorithm without
the implemented theoretical basis-provides the ground
systems advantages equivalent to those obtained by
integrating sensor engineering models onto the spacecraft
bus. Combined with a mature, repeatable process for
incorporating the theoretical basis, or scientific core, into
the "head" of the zombie algorithm, along with
associated scripting and registration, provides an easy
"on ramp" for the rapid and low-risk integration of
scientific applications into operational systems.
