Chapter 5
40
essential to the proper quantitative derivation of physical variables. The
requirements are not onerous since the atmospheric data need not be
acquired at spatial resolutions finer than 100 meters or so. In practice thus
far, however, this data is almost never acquired.
The same is true for the digital elevation model (DEM). Without
knowledge of the height and shape of the land surface, it is impossible to
adequately model the reflectance process or to correct in any accurate way
for atmospheric effects. The height of the land is necessary to derive how
much atmosphere the radiation being measured is passing through, and
hence how much correction to apply to each pixel once the atmospheric
model has been parameterized. The shape of the land is a critical factor in
applying corrections for the BRDF since they depend on the slope and aspect
of each pixel. The DEM is also critical to the accurate location of
measurements from space, in order to correct for relief displacement in the
case of measurements made by optical instruments, and layover distortion in
the case of measurements made with synthetic aperture radar both of which
require the height of the surface to be known.
The point made here is that since quantitative accuracy directly affects
the value of the information ultimately derived, the simultaneous acquisition
of atmospheric parameters and the availability of an adequate DEM are
critical factors influencing this value and hence the return on investment
achievable. A more detailed summary of atmospheric and geometric
corrections of Earth Observation data is to be found in MacDonald (1993).
What the data produced by the pre-processing step describes is what was
measured and the spatial distribution of that measurement. It tells us nothing
about what the measurement means. That is the purpose of the second step in
the process.
2.2
Data to information conversion
The second step we call Data to Information Conversion. In contrast with
the pre-processing step, which is driven entirely by the physics of the
measurement process and the technology, this second step is driven by the
information requirements of the operational user. In other words, the
processing methodology employed in the data to information conversion step
differs depending on the use to which the information is to be put. A simple
example will illustrate this point. Suppose we have measured the distribution
of spectral reflectance over a forested area. To the forester, this data can
provide input to a model which, when combined with other information, can
lead to conclusions about the health of the trees, the vigor of regrowth in
recently harvested areas or the onset of disease and things of that nature. To
the geologist, on the other hand, precisely the same data speaks of
40
essential to the proper quantitative derivation of physical variables. The
requirements are not onerous since the atmospheric data need not be
acquired at spatial resolutions finer than 100 meters or so. In practice thus
far, however, this data is almost never acquired.
The same is true for the digital elevation model (DEM). Without
knowledge of the height and shape of the land surface, it is impossible to
adequately model the reflectance process or to correct in any accurate way
for atmospheric effects. The height of the land is necessary to derive how
much atmosphere the radiation being measured is passing through, and
hence how much correction to apply to each pixel once the atmospheric
model has been parameterized. The shape of the land is a critical factor in
applying corrections for the BRDF since they depend on the slope and aspect
of each pixel. The DEM is also critical to the accurate location of
measurements from space, in order to correct for relief displacement in the
case of measurements made by optical instruments, and layover distortion in
the case of measurements made with synthetic aperture radar both of which
require the height of the surface to be known.
The point made here is that since quantitative accuracy directly affects
the value of the information ultimately derived, the simultaneous acquisition
of atmospheric parameters and the availability of an adequate DEM are
critical factors influencing this value and hence the return on investment
achievable. A more detailed summary of atmospheric and geometric
corrections of Earth Observation data is to be found in MacDonald (1993).
What the data produced by the pre-processing step describes is what was
measured and the spatial distribution of that measurement. It tells us nothing
about what the measurement means. That is the purpose of the second step in
the process.
2.2
Data to information conversion
The second step we call Data to Information Conversion. In contrast with
the pre-processing step, which is driven entirely by the physics of the
measurement process and the technology, this second step is driven by the
information requirements of the operational user. In other words, the
processing methodology employed in the data to information conversion step
differs depending on the use to which the information is to be put. A simple
example will illustrate this point. Suppose we have measured the distribution
of spectral reflectance over a forested area. To the forester, this data can
provide input to a model which, when combined with other information, can
lead to conclusions about the health of the trees, the vigor of regrowth in
recently harvested areas or the onset of disease and things of that nature. To
the geologist, on the other hand, precisely the same data speaks of