This chapter is a digression from the several
previous retrieval chapters. It deals with hardware design choices for
an MTP. During the "early days" of MTP work, which consisted of more
ground-based observing than airborne observing, it became evident that
since all retrieval procedures need a good quality
calibration the MTP design should rely upon
as much angle scanning as possible. As the JPL emphasis shifted to
airborne MTPs it slowly became evident that angle scanning was no more
prefereable than frequency sampling from the standpoint of
post-observation calibration. That is this chapter's "story."
In the previous chapters it was assumed that different altitudes were
"sampled" by observing at a selection of elevation angles with an MTP
having a small number of frequencies, such as 2 or 3 frequencies. Frequency sampling is an alternative strategy for achieving information from a wide range of altitudes.
During the "early days"
of ground-based microwave remote sensing of atmospheric temperature
profiles (the 1970s and 1980s) two groups were active, a JPL group (led
by Bruce Gary) and a NOAA Wave Propogation Labortory group (led by Ed
Westwater). The JPL group favored angle scanning and the NOAA group
favored frequency sampling. An argument can be made for each
philosophy. Angle scanning allows for the observation of a wide range
of applicable altitudes with a minimum of hardware, but it assumes that
temperature isopleths are flat and horizontal; frequency
sampling makes no assumptions about temperature isopleths and it
requires no moving parts. But frequency sampling requires a lot more
calibration than can be reasonably performed (at least that's my
position).
With angle scanning it is possible to inter-calibrate frequency
channels simply by comparing observables at elevation angles that have
the same applicable altitude. Provided isopleths are flat and
horizontal this inter-channel calibration should allow for the
adjustment of one or the other channel to achieve internal consistency.
For ground-based systems an average comparison during several days of
observations could be used to provide additional assurance that the
isopleths, on average, were horizontal. For airborne MTPs the solution
is simpler: merely compare channels while flying at different azimuths,
or make the comparisons when outside air temperature is not changing.
Both methods assure that the airborne MTP inter-channel calibration is
unaffected by non-horizontal isopleths.
The dynamic range of applicable altitudes that can be achieved is
limited by the antenna's beamwidth. A narrow beam (such as 5 degrees)
allows for a greater dynamic range than a wide beam (such as 10
degrees). This is due to the fact that the lowest applicable altitude
besides the horizon view is about a beamwidth above the horizon times
the sine of that angle times the applicable range.
Inter-channel calibration of an angle-scanning MTP usually requires
interpolation between observables. Note that sine all channels share
the same set of viewing angles (since they use the same horn antenna).
It makes sense to choose a set of viewng angles that provide uniformly
spaced "logarithm of applicable altitudes" (to minimize observable
redundancy), and it is unlikely that two frequencies will have the same
applicable height for any pair of viewing angles. Therefore,
inter-channel calibration usually requires an interpolation between
observables that straddle the applicable altitude of the other channel.
As a practical metter this is not a problem.
For an airborne MTP an alternative procedure for achieving calibration of each channel has become common. The
airborne MTP requires a fairing and a window with substantial strength,
and high density polyethylene is normally used (because of it's low
dielectric properties). For each channel absorptions and reflections
are likely to be different for different viewing angles. This is partly
due to the ray path through the window being slightly different for
each viewing angle since the window has to conform to the shape of the
fairing and the fairing shape is usually not a section of a circle for
the entire range of viewing angles. Another effect could be more
important, and that's differences in dirtiness of the window material
between locations. It is common practice to use a stiff bristle brush
to clean the outside of the fairing before each flight in order to
minimize changes in absorption or reflection at differenet locations on
the window. (The window is grooved on the outside and inside surfaces
to provide a "lens coating" effect to minimize reflections, and these
grooves are better sites for dirt accumulation than a smooth surface.)
There is a lingering concern that the directional coupler may not
prevent LO singal from reaching the horn where it would then reach the
window and be partially reflected back into the horn to create a
standing wave radiometer output level offset. This offset would vary
with viewing angle since the distance between the horn and the window
is not exactly the same for each angle. If this leaking LO standing
wave offset is present it could cause an additional component of
required angle dependent calibration.
For these various reasons it is unwise to attempt inter-channel
calibration using an airborne MTP. Instead, each channel ant each
viewing angle is calibrated by comparing TB observables with predicted
ones based on RAOBs. Whenever the MTP flies close to a RAOB site this
"window correction table" (WCT) can be computed. Typically 5 to 10 such
WCTs are required to achieve sufficient accuracy. A median at each
element of the WCT ensemble is better than an average, since temporal
and spatial interpolations of RAOB T(z) profiles can sometimes be
innacurate.
As the previous two paragraphs should have made clear, calibration of
the airborne MTP is no simpler using angle scanning than frequency
sampling. Thus, it should be left to the hardware engineer whether to
use one or the other, or how to partition reliance upon the two
approaches. The engineer's goal should be to maximize the range of
applicable altitude sampling in the shortest time for completing an
observing cycle. So far the airborne MTPs that JPL has flown have
included 2 or 3 frequencies that are selected for each of 10 elevation
angles.
Go to next Chapter #12
Go to previous Chapter #10
Return to Introduction
____________________________________________________________________