U.S. Department of Energy

Pacific Northwest National Laboratory

Identifying Cloud Phase States from Multiple Remote Sensing Observations

Cloud feedbacks remain one of the largest uncertainties in climate models and contribute to key modeling errors in atmospheric processes like dynamics and precipitation. A more quantitatively rigorous identification of phase from remote sensors can improve process model simulations of the radiative effect of clouds, leading to improved accuracy of parameterizations in global climate models.


We have measurements from two active sensors at multiple Atmospheric Radiation Measurement sites with the potential for identifying cloud phase. However, not all the available information is currently used in cloud phase identification algorithms, and the algorithms lack statistical rigor in their implementation and quality assessment.

We are using Trelliscope visualizations, statistical clustering algorithms, and expert knowledge as tools to explore what new information is available in higher moments of the radar Doppler spectra. In the future, we plan to use Bayes net classifiers and Signature Quality Metrics to build algorithms with operational knowledge of signature identification performance given the subset of measurements available in a particular case.


Results from early research shows potential for improving cloud phase by using additional information from radar Doppler spectra. These results were presented at the American Meteorological Society Cloud Physics conference in July 2014.

Project Staff: 
Laura Riihimaki, Jennifer Comstock
| Pacific Northwest National Laboratory