The in situ lidar detector housing on the left wingtip tank of the SPEC operated Learjet.
The in situ lidar detector housing on the left wingtip tank of the
SPEC operated Learjet.
The detector module with electronics boards and detector head. The
electronics outputs high gain and low gain channels at 50 MHz sampling
rate and implements amplifier gain switching to provide up to 6 orders
of magnitude dynamic range. The electronics also automatically adjusts
the photomultiplier tube (PMT) gain.
The detector head with upward and downward viewing PMT detectors. Each
detector has wide field-of-view (30o half angle) optics for
nighttime operation and narrow FOV (3o) optics with a 0.37 nm
(FWHM) solar blocking filter for daytime operation.
The laser in the Learjet cabin. The 532 nm wavelength YAG laser fires
180 mJ per pulse at 10 Hz through an optical flat in a cabin window on
the right side of the plane.
An example lidar signal in dense cloud on the first engineering flight
during daytime. The calibrated lidar data from the low and high gain
channels is shown with the red and blue lines. The green line shows the
data from the two channels merged and averaged over log spaced time bins
and with the solar background signal subtracted. The error bars show
the standard error of the photon fraction from the variability in each
time bin. In this example with the daytime optics, the peak lidar
signal is about 40 times above the solar background. With background
subtraction the dynamic range is more than three orders of magnitude and
the usable signal extends beyond 15 microseconds. The solar zenith angle
was 76o at this time.
The path of the cloud portion of the science flight on 2005-12-02 UTC
off the coast of south Texas. The dots along the line indicate when the
aircraft was in cloud as indicated by an FSSP concentration of more than
10 cm-3. The numbers along the flight track are the times in
UTC hours.
An example of the calibrated, merged, and time bin averaged lidar
signals for up and down detectors from the science flight. The fits of
the function, log[p(t)] = a - b*log(t) - ct, are shown and the
coefficients listed in the legend. These three fit coefficients for
each detector are input to a neural net (trained on simulated in situ
lidar signals in stochastic stratocumulus clouds) to retrieve extinction
at four averaging volume sizes, cloud thickness, and cloud relative
aircraft altitude.
Examples of the lidar signal at five selected photon times from 0.5 us
to 8.0 us as a function of laser shot time for the up and down
detectors. The signal sampled at later photon times changes much more
slowly with the aircraft travel than the signal at shorter photon times,
due to the larger volume sensed by the diffusing photons.
A scatter plot of the lidar retrieved extinction versus the FSSP
derived extinction. The 1:1 and 1.8:1 lines are shown. The linear
correlation in log extinction of the points is 0.915. This correlation
is quite high considering that the lidar is measuring a cloud volume
more than 1010 times larger than the FSSP probe in an
inhomogeneous cloud. The offset of the lidar extinction from the FSSP
extinction is a factor of 1.8, which is mostly due to the uncertainty in
the lidar calibration.
Cloud base and top altitude derived from lidar retrieved cloud
thickness and cloud relative altitude and the aircraft altitude. Also
shown are the aircraft altitude and the top and base altitudes obtained
from cloud boundaries derived from the FSSP probe. The vertical dotted
lines indicate the times of furthest south or north latitude. The lidar
retrieved cloud base and top altitude shows a consistent trend to lower
altitudes to the south and higher altitudes to the north. There is
reasonably good agreement between the lidar retrieved and aircraft
derived cloud boundaries, though there is considerable variation in
the aircraft derived boundary altitudes.