HR: 1340h
AN: SF43A-0776 [Abstracts]
TI: Remote Sensing Opportunities Beyond Exploration of the Photon's State Variables, with
Examples
AU: * Davis, A B
EM: adavis@lanl.gov
AF: Los Alamos National Laboratory, Space & Remote Sensing Sciences Group, Los Alamos, NM 87545
AU: Cox, L
EM: larrycox@lanl.gov
AF: Los Alamos National Laboratory, Space & Remote Sensing Sciences Group, Los Alamos, NM 87545
AU: Funsten, H
EM: hfunsten@lanl.gov
AF: Los Alamos National Laboratory, Center for Space Science and Exploration, Los Alamos, NM 87545
AB:
The evolution of Earth system observation by satellite remote sensing can be seen as a systematic exploration and
exploitation of the photon or EM wave's state variables: emission (or last scattering) point and direction, wavelength, and
now polarization, plus travel time in active techniques. Wavelength spawns the spectral dimension of the data, itself
evolving from broadband to multi- to hyper-spectral sampling and to ever more exotic wavelengths (e.g., microwaves and sub-mm
wavelengths). We are thus constantly pushing the technology to open new spectral ``windows'' or to examine the available
ones in increasing detail. Polarization, a statistical property of the photon beam, offers promising applications. Position
and direction are geometrical beam characteristics that simply determine the pixel of interest on the focal plane; no more
is expected. Similarly the temporal dimension in active techniques is interpreted just as return-travel time, hence position
along the source beam. Conventional exploitation algorithms (i.e., retrieval schemes) look for known patterns, trends,
correlations, etc. between radiances at different wavelengths and/or polarizations. This strategy seems natural and enables
independent pixel-by-pixel exploitation, a significant advantage for the data processing as the numbers of pixels and
spectral/polarization channels increases. However, at this level, we are overlooking all the information that could be
reaped from the complex spatial/directional relations that are so obvious when we examine satellite imagery visually. At
best, this is left for projects in post-processing that tend to be limited to case-studies published in the research
literature.
We will argue that this strategy of extensive photon-state sampling and the resulting ---or is it driven by?---
pixel-by-pixel exploitation paradigm in remote sensing is suboptimal for a number of reasons. For one, it comes hand-in-hand
with making horizontal homogeneity assumptions inside the pixel footprint that may not be realistic enough to approximate the
actual radiative transfer processes at the accuracy of the instrument's SNR level. Furthermore, as spatial resolution
increases, there will be increasing amounts of radiometric cross-talk between pixels (a.k.a. adjacency effects) at
scattering/reflecting wavelengths; this impact of spatial variability will also be ignored. A cost-effective remedy for the
ills caused by these internal and external variability effects is to engage the three-dimensional radiative transfer modeling
community in improved algorithm development.
Last but not least, there is an emerging class of instruments that are designed from the onset using the rich phenomenology
of 3D radiative transfer, often with time dependence; the theory that supports these instruments is not so much about
individual photon beams but about the complex interaction of the photon population as it flows within the 3D
atmosphere-surface medium. The examples we will use include high-resolution O$_2$ spectrometry as a cloud-scene probe and
off-beam cloud lidar systems. Because of the radically different kind of theory involved, these instrumental developments
are not evolutionary but revolutionary. This poses a challenge for mission and program management as well as for the
developer. How does one promote a promising but unconventional instrument design into an operational setting?
UR: http://nis-www.lanl.gov/~adavis
DE: 5494 Instruments and techniques
DE: 6919 Electro-optics
DE: 6944 Nonlinear phenomena
DE: 6969 Remote sensing
DE: 3360 Remote sensing
SC: Special Focus: Advances in Data Acquisition, Management, Analysis and Display [SF]
MN: 2004 AGU Fall Meeting