Detection and Retrieval of Cirrus Clouds in the Tropics from AIRS: Validation from ARM Data

Yue, Q., University of California, Los Angeles

Cloud Distributions/Characterizations

Cloud Properties

Yue Q and KN Liou. 2009. "Cirrus cloud optical and microphysical properties determined from AIRS infrared spectra." Geophysical Research Letters, 36, L05810, doi:10.1029/2008GL036502.


Sensitivity of the sub-band BT spectra to variation in optical depth and mean effective diameter for optical depths of (a) 0.1, (b) 0.5, (c) 1.0, (d) 2.0, (e) 3.0, and (f) 4.0. Surface and cloud temperature were set to be 300 K and 200 K, respectively. Circles indicate the position of each sub-band. Black, gray, and light gray solid lines indicate a cirrus layer with mean effective diameters of 30, 57, and 92 mm, respectively.


Four AIRS thin cirrus cloudy BT spectra calculated from simulation and observation in the 750–1130 cm-1 region. The cloud properties were determined by a minimization method.

Composed of nonspherical and irregular ice crystals with sizes ranging from a few to thousands of microns, a direct satellite determination of the cirrus microphysical and optical properties of ubiquitous high tropical cirrus clouds (9-18 km) is an involved and challenging task.

In the past few years, the atmospheric infrared sounder (AIRS), a high-spectral-resolution sensor in the thermal infrared spectral region, has provided rich information content for the determination of a variety of atmospheric parameters. In addition to strong CO2 and H2O bands for temperature and humidity profile retrievals, AIRS also has thermal IR window channels that are particularly useful in inferring surface and cloud properties for both daytime and nighttime. In our previous report, we presented a fast radiative transfer model that has been applied to AIRS window spectra for retrieval of the optical and microphysical properties of thin cirrus with optical depths less than 0.3 (Yue et al. 2007), which neglected multiple scattering contributions for cloud particles.

We have developed an efficient thermal infrared radiative transfer model on the basis of the delta-four-stream approximation to facilitate application of AIRS data to account for multiple scattering contributions from cirrus particles, significant at optical depths larger than 0.3. Numerical experiments demonstrated that information content in the 800-1130 cm-1 thermal infrared window spectral region is sufficiently distinct for the inference of cirrus optical depth and ice crystal mean effective size and shape factor. We analyzed 312 nighttime cirrus pixels in two AIRS granules over ARM Climate Research Facility (ACRF) Tropical Western Pacific (TWP) sites and applied the radiative transfer model to these cases to determine cirrus optical depth and ice crystal mean effective size, based on a look-up table approach. The retrieval program has been evaluated through an error budget analysis and validation effort by comparing AIRS-retrieved results with those determined from ground-based millimeter-wave cloud radar (MMCR) data at ACRF TWP sites, collocated and coincident with AIRS overpasses (Yue and Liou 2009).

This infrared radiative transfer model has been successfully applied to retrieve cirrus optical depth and mean effective size from the AIRS spectra covering 13 "clean" sub-bands in the 10 mm region of the spectra. We have conducted a number of case studies and selected 312 tropical cirrus cloudy pixels from two AIRS granules collocated and coincident with the mm-wave cloud radar measurements over the ACRF TWP sites for analysis and inter-comparison. Using the retrieved cirrus parameters, we have shown that the cloudy BT spectra in AIRS subbands simulated from the infrared radiative transfer model deviate from the observed spectra by less than 0.6 K. Moreover, we illustrated that the cirrus optical depth and mean effective ice crystal size retrieved from AIRS spectra are generally consistent with those derived from the ARM MMCR reflectivity results.

Our analysis shows that the sensitivity of infrared BTs to ice crystal habit appears to be smaller than that due to the effects of optical depth and ice crystal size in the 800–1130 cm -1 window spectral region. We also investigated retrieval errors associated with the uncertainty and instrument noise in AIRS measurements. Uncertainties in temperature, humidity, and cloud top temperature are the major error sources in the retrieval; however, these uncertainties are much smaller compared to the BT sensitivity to optical depth and ice crystal mean effective size.

The radiative transfer and remote sensing algorithm presented in this paper can be applied to AIRS radiances for the inference of cirrus cloud optical depth and mean effective ice crystal size over the tropical region and—with modification—can be applied to other regions of the globe as well.