Figure 31 shows the probability 
as a function of time and altitude
for quarter-hourly averaged wind measurements on July 26, 1989.
The contours show the wind speed
increasing from 5 ms
to 8.5 ms
over the period.
Even using 0.2 ms
contour spacing, the wind profiles are smooth
and small details are visible.
The measurements have good internal consistency, except
for some small disturbances with a smaller 
.
The results are most reliable in the convective boundary layer,
since there is good contrast between
convective aerosol structures and clearer free air.
Above the convective boundary layer there are sometimes not enough
aerosol structures for correlation analysis, which appears as
less reliable results.
Turbulent mixing in the late afternoon
frequently produces uniform aerosol distribution
that leads to less reliable wind estimates.
In the convective boundary layer, randomly located
smaller reliabilities are generally due to the breaks in the VIL data record.
This happened several times on July 26 after 14:00 CDT.
These breaks are generally due to system adjustments or
safety shutoff of the laser output beam when an aircraft flew
into the field of view.
Figure 31:
Quarter-hourly VIL wind speeds (contours), mean boundary layer (solid line),
and cloud base (dash-dotted line) plotted as functions of
time and altitude on July 26, 1989.
The shaded areas indicate 
is less than 1.
In general,
the results are reliable from the ground to 500 m above the cloud base.
There are gaps in the data record at about 14:00, 16:00,
and 17:00 CDT causing 
to decrease.
Boundary layer mixing decreases the aerosol structure
contrast in the volume scans and occasionally causes less reliable estimates
in the middle of the mixed layer at 13:00, 14:40, and 15:30 CDT.