next up previous
Next: AERI Microwindow Regions Up: Abstract and Contents Previous: AERI and HSRL Derived Brightness Temperature

Return to the Publications. Return to the Index.


An increased understanding of cirrus clouds is necessary to improve current climate models. This includes atmospheric and environmental factors which drive the formation of cirrus, database maintenance of regional and seasonal cirrus coverage on a global scale, and the effect of cirrus on radiative transfer calculations. The research discussed presently focused on the radiative effect of cirrus and, if continued, would lend to a seasonal analysis.

Independent remote sensing at visible and infrared wavelengths, using ground based high spectral resolution lidar (HSRL) and interferometry (AERI), respectively, produced cirrus cloud optical depth and atmospheric column brightness temperature measurements for Madison, WI. A series of 19 infrared microwindows, spectral regions between water vapor absorption lines, were compared to visible lidar measurements. Local atmospheric temperature and dewpoint temperature information was acquired by releasing a radiosonde. Clear sky radiance was calculated using the FASCOD3P line-by-line transmission model.

Trends in the measured optical depth and downwelling brightness temperature were consistent. However, differing instrument field of views and atmospheric dwell-times are limiting features and apparent in portions of the data. Despite these limitations, the HSRL derived brightness temperatures compared favorably with AERI measurements, typically within 5 K over a 100 K temperature range. Visible and infrared optical depths also exhibited good correlation. The HSRL has an upper limit visible optical depth measurement bound of approximately 3.0, defined by the transmitter and detector capabilities. Similarly, the upper bound of AERI calculated infrared optical depth occurs between 2.5 and 3.0; where small deviations in the cloud radiance yield large changes in the optical depth.

Radiosonde and FASCOD3P model uncertainties for infrared data and multiply scattered return for visible data also introduced an important error contribution. Atmospheric water vapor and aerosol loading within the first few kilometers above the surface introduced a spectral bias to the AERI measured radiance. This affected the minimum resolvable optical depth that could be measured by the AERI, and was corrected by scaling the clear atmospheric radiance and transmittance accordingly. The FASCOD3P calculated clear sky radiance can yield significant errors in the data inversion if not corrected relative to AERI measured clear sky data. The resulting difference between FASCOD3P and AERI radiances was shown to account for over 50 percent of the total column radiance for optically thin cirrus and clear conditions. This produced greater than 7 K differences in the derived brightness temperatures. Overall, spectrally dependent radiance errors ranged from 1 to 8 percent of the total cloud radiance. This corresponded to uncertainties in brightness temperature of 0.5 to 3 K for uniform transmissive cloud cover. An approach was given to estimate the brightness temperature for opaque conditions, resulting in a 10 K difference between the transmissive and opaque calculations. The radiance errors yielded a maximum infrared optical depth deviation of 0.1 for a measured value of 0.9. Multiple scattering was shown to account for as much as 20 percent of the HSRL aerosol return signal, which resulted in an underestimation of the HSRL measured visible optical depth.

The visible to infrared optical depth ratio, tex2html_wrap_inline3377, results were determined by two separate approaches: calculate the visible and infrared optical depths, then determine the ratio; and calculate the downwelling radiance using the HSRL visible optical depth, while iterating the optical depth ratio until the radiance matched the AERI value. Analysis of the unweighted data sets required constraints for optically thin (visible OD < 0.1) and thick (visible OD > 2.0) cases. Iteration of tex2html_wrap_inline3383 for the weighted cloud was shown to be inconsistent below a visible optical depth of 0.5. Data below this value was not used in the spectral analysis.

Iteration of the optical depth ratio was expected to achieve superior results because the derived values are based on a weighted cloud extinction cross-section, and the data from each instrument is coupled. The other approach assumes a uniform extinction cross-section throughout the cloud, where the optical depth solution is instrument independent. These techniques can be extended to calculate the upwelling radiance which would be measured by space-borne instrumentation by applying a top-down solution of the radiative transfer equation, assuming the atmosphere does not become opaque due to optically thick cloud cover.

The weighted approach yielded a decreased scatter in the tex2html_wrap_inline3385 results, which indicated a small bias relative to Mie theory. It was concluded that the direction of the bias was due to instrumental field of view and dwell-time differences, which dominated the combined multiple-scatter and FASCOD3P errors. Implementation of a telescope on the AERI would improve the field of view difference between the instruments. This would increase the light gathering power of the AERI, thereby allowing a minimization in the field of view given similar data acquisition times. Spatial and temporal errors are inherently coupled as the atmosphere is advected through the instrumental field of view. The current approach does not account for AERI calibration times when comparing the AERI and HSRL data. Limiting the HSRL data to AERI atmospheric view times would further improve the instrument correlation.

Particle size spectral dependence was also a factor and the data was shown to agree with Mie theory for ice spheres. A spectral region with a minimum particle size sensitivity near 920 cmtex2html_wrap_inline3387  was suggested by Mie theory calculations. The optical depth ratios for the combined cases were spectrally similar to 35 tex2html_wrap_inline3389m radius (50 tex2html_wrap_inline3391m assuming a uniform cloud) ice spheres.

The measured visible to infrared cirrus cloud optical depth ratio can be used in climate models. A parameterization over the entire infrared atmospheric window would require a mean visible to infrared ratio of 1.9 based on 50 to 100 tex2html_wrap_inline3393m radius particles. This value could be scaled based on Mie theory results if a mean particle size of the cirrus could be obtained. The large change in ice absorption across the infrared spectrum resulted in a particle size dependence, which could be important to climate models for smaller particles. As the particle size decreased, there was a greater dependence on particle size across the infrared window. The visible to infrared optical depth ratio varied from 1.9 (for 100 tex2html_wrap_inline3395m radius ice spheres) to 2.5 (for 25 tex2html_wrap_inline3397m radius ice spheres) between 1000 and 1150 cmtex2html_wrap_inline3399. However, analysis of ice crystals smaller than 25 tex2html_wrap_inline3401m radius should be performed for validation to the Mie theory ice sphere results.

Observation of contrails formed as a result of jet engine exhaust would allow study of smaller sized particles, relative to cirrus cloud particles. A case study similar to the one presented in this thesis for cirrus would provide a spectral comparison of visible to infrared optical depths. This could further confirm the particle size dependence suggested by Mie theory for ice spheres. It would also show the radiative impact of jet engine exhaust pollution near large airports.

The acquisition of additional data would be useful to complement the current data set, and to improve the statistical analysis relative to Mie theory. A total of 7 of the 8 data sets given in Table 3 were useful in the analysis. The remaining case, 2 December 1995, was useless for cirrus study due to an opaque water cloud below the cirrus cloud layer. The data inversion process fails for water cases because the optical depth of water droplets is large relative to ice crystals. This results in a large optical depth gradient over 200 to 300 m. The current algorithm is based on the FASCOD3P vertical resolution of 64 data points, which yields a layer thickness of 100 to 500 m for a strong to weak dewpoint temperature gradient, respectively. This implies that the depth of the water cloud deck would have the proper resolution. Nonetheless, it is a problem that can be resolved with additional cases and is beyond the scope of this thesis, which is focused on cirrus clouds.

Errors associated with FASCOD3P radiance and transmittance calculations within the water vapor continuum might be decreased by utilizing the high spectral resolution data sets. HSRL measured aerosol backscatter cross-section and depolarization, coupled with AERI radiance measurements, could provide the information required to yield a vertically weighted atmospheric transmittance to correct for the FASCOD3P column radiance relative to AERI measured values. This would improve the current approach, which uses the Gamma correction to apply a uniform correction to the transmissivity.

next up previous
Next: AERI Microwindow Regions Up: Abstract and Contents Previous: AERI and HSRL Derived Brightness Temperature

Daniel DeSlover
Sun Aug 11 10:02:40 CDT 1996