ARM Sites Enable Assessment of Cluster Analysis for Identifying Cloud Regimes

Jakob, C., Monash University

Cloud Distributions/Characterizations

Cloud Properties

Jakob, C., and G. Tselioudis, Objective identification of cloud regimes in the Tropical Western Pacific, Geophys. Res. Lett., 30(21), 2082, doi:10.1029/2003GL018367, 2003.

To diagnose and improve the treatment of cloud radiative effects in general circulation models, researchers supported by DOE's Atmospheric Radiation Measurement (ARM) Program are evaluating techniques that identify cloud types by various groups, or regimes. Different types of clouds produce different radiative effects, depending on cloud properties like optical depth and particle size, as well as environmental conditions such as temperature. Grouping clouds by these properties and then adding in other data such as relative frequency of occurrence and precipitation provides important representative information that can be used for a wide range of purposes, from model evaluation to planning focused field experiments.

Research reported in Geophysical Research Letters (Nov. 2003) shows the results of a technique aimed at identifying cloud regimes based on observed cloud information. A basic question in this study was whether observed cloud fields were composed of a collection of random cloud situations, or of distinct and recurring cloud regimes. Using satellite data obtained during 1999 by the International Satellite Cloud Climatology Project, joint histograms of several cloud parameters in the Tropical Western Pacific (TWP) were developed. The researchers then applied an iterative clustering algorithm to converge on like conditions, or clusters, in the data.

Four major cloud regimes emerged from the cluster analysis: a shallow cumulus regime; a cirrus regime with thin clouds; a regime of thicker cirrus associated with convection; and a deep, probably organized, convective regime. With this information, the researchers were able to conduct a quantitative assessment of the relative role of each of the regimes through measurements from ARM locales on Manus and Nauru islands in the TWP. Using the 1999 histograms from the corresponding ISCCP grid box containing each site, about 1,500 cloud situations from each site were analyzed. All four cloud regimes identified by cluster analysis for the TWP did occur over the ARM sites, with different frequencies of occurrence related to specific site location (sea surface temperature distribution in the region).

Cluster No. TWP Manus Nauru
1 46% (0.40) 34% (0.33) 73% (0.26)
2 23% (0.90) 21% (0.91) 10% (0.85)
3 17% (0.77) 33% (0.76) 16% (0.68)
4 14% (0.98) 12% (0.99) 1% (0.98)

Relative frequency of occurrence and total cloud cover (in parentheses) for the four TWP cloud regimes for the entire TWP and for individual ISCCP grid boxes located around the ARM sites at Manus, PNG, and Nauru Island.

For the TWP, the cluster analysis method proved useful for robustly identifying physically interpretable cloud regimes. The subsequent quantitative assessment using ARM's Manus and Nauru sites allowed an evaluation of the representatives of those sites for the cloud regimes occurring in the wider TWP area, while also providing context for specific cloud regime studies at those sites. As the study demonstrates, cloud regime identification enables the placement of individual measurement sites into a large scale context. This allows for a more quantitative generalization of the results of detailed process studies with respect to their role in climate. Through similar measurement data available at other ARM locales, the analysis of additional geographical areas and longer time periods will help to establish a global picture of the distribution and frequency of cloud regimes, and their role in radiative feedback processes.

ARM observations will enable the further characterization of the radiative, cloud, and thermodynamic properties of the TWP cloud regimes. Work in this area has been completed and will be reported in a publication in the near future.