Cirrus clouds were viewed using the University of Wisconsin Volume Imaging Lidar, University of Wisconsin High Spectral Resolution Lidar, and the VAS radiometer situated on GOES. The VIL imaged the cirrus clouds within a mesoscale volume. The HSRL measured the visible scattering properties of the cirrus clouds. The VAS radiance measurements were used to calculate the infrared absorption properties for the cirrus clouds.
The backscattered radiation detected by the VIL was
used to determine the horizontal and vertical structure of
the cirrus clouds. The presence of cirrus clouds was determined by
choosing a threshold value from a histogram of the
VIL backscattered signal between heights of 6 and 11 km.
The backscattered radiation in each VIL profile was compared to the
threshold value to determine cirrus cloud cover percentages and structure
functions. The cirrus cloud cover percentages ranged from 54.7%
to 100% for simulated vertical profiles across the wind created from the
VIL cross wind scans. For the two time periods studied, the area averaged
cirrus cloud cover percentages were 81.5% and 76.8%.
Differences in the average cloud cover were seen between the VIL area
measurements and HSRL vertical profiles. The maximum cloud cover
difference of 45.3% between two VIL RTIs during the first time period
was used as a cirrus cloud cover error to estimate a resulting
change in the incoming solar radiation. This cloud cover error, along
with an average cirrus cloud optical depth of 0.257, resulted in an
estimated change in the direct solar flux
of 58.18 W m
seen at the surface of the Earth. For the second time
period, a cloud cover difference of 8.9% between the VIL and HSRL,
associated with a cloud optical depth of 0.428, resulted in a 12.36 W m
difference in direct solar flux
seen at the surface of the Earth. The errors due to the incorrect
cirrus cloud cover would dominate the increase in the planetary effective
temperature resulting from the doubling of CO
, especially in the case
of spatially scattered, optically thin cirrus clouds. This
implies that the variability of the cirrus clouds throughout a
mesoscale volume has to be measured to understand the effects of these clouds
on the Earth's radiation balance. Sampling errors associated with point
measurements make them poor indicators of the cirrus cloud area averaged
values.
The cirrus cloud horizontal and vertical structure was also examined. Structure functions were used to determine the cirrus cloud length, width, and the separation between clouds. The cirrus cloud average length along the wind was 130 km and its length across the wind averaged 14 km. The average distance between clouds along the wind was 273 km while across the wind it was 24 km. In this experiment, the cirrus clouds were typically aligned along the wind direction with an aspect ratio of approximately 9:1. Preferential alignment of cirrus clouds implies that point measurements which rely on cloud advection may not even detect cirrus clouds if large variations exist across the wind. Examination of the vertical cloud structure showed many instances of multi-layered and/or precipitating cirrus clouds. The complexity of the vertical structure shows the dependence of cloud formation on the dynamical situation in the atmosphere.
A method was described to calculate the cirrus cloud visible optical properties across a mesoscale volume. This was possible because an experiment was run where the VIL along wind scan plane crossed over the HSRL position resulting in coincident cirrus clouds measurements. The HSRL cirrus cloud aerosol backscatter cross sections were used to directly calibrate the VIL backscattered signal. A point comparison was made for the HSRL aerosol backscatter cross sections and the VIL data (which was normalized to a low level aerosol layer to remove angular dependencies). This initial comparison was computed for the signal backscattered from the bottom 1.5 km of the cirrus clouds where it was assumed that the attenuation in the VIL signal was negligible. A best fit straight line was used to calibrate the VIL backscattered signal into aerosol backscatter cross sections per unit volume. The calibrated VIL signal was then attenuation corrected using a forward integration of the single channel lidar equation. To forward integrate the lidar equation, extinction cross sections at each data point and a multiple scattering correction factor were needed for the VIL data. Extinction cross sections were created from the VIL calibrated aerosol backscatter cross sections using a single aerosol backscatter phase function for the whole volume calculated from the HSRL data (assuming negligible molecular extinction within the cirrus cloud compared to the aerosol extinction at 1064 nm). A multiple scattering correction factor of 0.5 was used to correct the backscattered signal to account for one half of the original scattered light being diffracted by the ice particles and staying within the receiver field of view. After the VIL signal was attenuation corrected, the resulting calibrated VIL cirrus cloud aerosol backscatter cross sections were compared to the original HSRL aerosol backscatter cross sections. An adequate correlation resulted between the two sets of aerosol backscatter cross sections. The largest errors in the calibration technique resulted from VIL scan angle errors leading to misalignments between the two vertical profiles, misalignments between the VIL scan direction and the wind direction, and the usage of a bulk aerosol backscatter phase function in the VIL attenuation correction technique. The alignment of the VIL scan and the HSRL vertical profile was critical since the calibration technique was dependent upon both system viewing the same cirrus cloud.
The calibration technique was used to convert the VIL signal in both scan directions into aerosol backscatter cross sections per unit volume. The calibrated cross wind VIL data was used to calculate the visible scattering optical depth of the cirrus clouds within the mesoscale volume assuming no changes in the cirrus clouds as they were advected by the wind (over a half hour period). The visible optical depths were calculated by integrating the VIL cirrus cloud extinction cross sections along a path through the VIL observed volume traced by a ray from the position of the GOES satellite. This allowed for a direct comparison of the VIL visible scattering optical depths and VAS infrared absorption optical depths. The ratio of the VIL visible scattering optical depth to the VAS infrared absorption optical depth was approximately 2:1 (especially for thin cirrus clouds) although variations did occur. The 2:1 ratio value agrees with previous measurements by Minnis (1990). When making this optical depth comparison, the cirrus clouds detected by the VIL had to be correctly allocated into the different pixels to enable an accurate comparison with the infrared image. This process was hampered by the lack of variation in the infrared image compared to the visible cirrus cloud image.
The technique to compare the visible scattering to infrared absorption
optical depths can
also be used to compare the cloud cover determined from the VIL to the
cloud cover calculated from the satellite infrared radiometer data using the
CO
slicing technique. This comparison would be used to test
the accuracy of the satellite based cirrus cloud climatologies.
The cirrus cloud albedo and mid-cloud height calculations from satellite
based radiometers can also be tested. The level of maximum scattering
within a cirrus cloud can also be determined and compared with the cirrus
mid-cloud heights throughout the mesoscale volume. Although
these comparisons are not encompassed in this thesis they
can easily be accomplished with the tools available.
This first attempt at a calibration of the VIL backscattered signal by the HSRL aerosol backscatter cross sections shows promise. Some of the inherent problems in this technique were revealed. Improvements are being made to both the VIL and the HSRL. This will result in more accurate measurements which will lead to a better understanding of the cirrus cloud optical and structural properties. The cirrus cloud detection technique and the VIL calibration technique should be attempted on a large cirrus cloud data set to achieve a statistical representation of the cirrus cloud optical and structural properties.