Determining radiative heating profiles from satellite data using statistical classification and cluster analysis

Knowledge of the vertical structure of radiative heating rates is important to understand the distribution of energy within the atmosphere. Currently, detailed information on vertical radiative heating profiles can be only be derived from instruments measuring vertical cloud property profiles in the atmosphere, such as the radar and lidar instruments at Atmospheric Radiation Measurement (ARM) program sites. However, these measurements typically represent horizontal scales much smaller than in climate models, making it difficult to compare the two sets of cloud and radiation fields. By modeling the relationship between the ARM surface based measurements and geostationary satellite measurements, knowledge of the vertical structure from ARM measurements can be expanded to a larger horizontal scale, giving a better understanding of the full atmospheric radiation budget.

In previous work, states of radiative heating rate profiles were defined, based on thresholded bins of observed atmospheric variables. We expand these state definitions using statistical clustering techniques to identify distinct patterns of cloud and heating rate profiles present in the observational data. Validation is done by comparing the resulting per-state distributions of independent atmospheric observations to known physical conditions. After defining a stable set of states, we build a classification model to link observed geostationary satellite data to the set of defined states. With the classification model, vertical heating rate profiles can be calculated from the satellite data without direct ground based observations.