Diagnosing cloud occurrence biases in the AM3 at SGP using atmospheric classification

Description

We define a set of atmospheric states for a region surrounding the ARM Southern Great Plains (SGP) site using an automated clustering technique. Atmospheric state in this context can be thought of as a frequently occurring regional weather pattern. We define states using ERA-Interim reanalysis and use cloud occurrence data from the cloud radar at the SGP site to validate the statistical significance of each state. We then classify the state of the atmosphere every 6 hours for the duration of the study, creating a time series of atmospheric state. This time series is used to composite ISCCP data to create joint cloud top pressure – optical depth histograms of cloud occurrence for each state. Snapshots of output from the AM3 model are sorted according to the observed atmospheric states, and their associated distributions of cloud occurrence from the ISCCP simulator are composited to produce modeled joint histograms of cloud occurrence for each atmospheric state. Comparison of the observed and modeled distributions of cloud occurrence for an individual state provides a test of the model parameterization under particular sets of physical conditions. In contrast, comparing the observed frequency of occurrence of the atmospheric states with their occurrence within AM3 tests how well the model reproduces the conditions of the region. Doing so allows us to parse the model’s total bias of cloud occurrence into contributions from the parameterization, and contributions from the distribution of states. We find that the model lacks high thin cloud under all conditions in the model, but that biases in deep thick cloud are state-dependent. We show that frontal conditions in the model do not produce enough deep thick cloud, while weather patterns associated with isolated convection produce too much. Increasing the horizontal resolution of the model improves both of these biases, but through different mechanisms. The high resolution run also changes the distribution of states however, ultimately increasing the total cloud occurrence bias. This creates an interesting question of whether or not the high resolution run performs better than the low resolution.