From Measurements to Models: Cross-Comparison of Measured and Simulated Behavior

This project establishes a update methodology for objectively comparing Atmospheric Radiation Measurement (ARM) data to the outputs of leading climate models and reanalysis data. The quantitative basis for this comparison is provided by a statistical procedure, which establishes an exhaustive set of mutually-exclusive, recurring states of the atmosphere from sets of multivariate atmospheric and cloud conditions, and then classifies multivariate measurements or simulation outputs into those states. Dynamic behavior of the atmosphere, whether measured or simulated, consists of a changing time sequence of such states. If models and measurements agree, the multivariate conditions from each source should classify the atmosphere into the same chronological sequence of atmospheric/cloud states. Comparing ARM data with the models in this way will establish a blueprint for achieving ARM's goal of improving models with derived relationships observed in the measurements. Such statistical comparison techniques will close the measurement-model loop by producing useful feedback for model developers. In doing so, this project will drive the need for additional data collection and lead model developers to deliver improved models supporting policy makers in the determination of acceptable levels of greenhouse gases in the atmosphere. Meeting this objective directly supports the Biological and Environmental Research Climate Change Research Division's Long Term Measure of Scientific Advancement.