Objective measures of the information density of satellite data

We propose and objective method whereby the density of Shannon's “information” associated with the retrieval of a profile of atmospheric variables from satellite-derived infrared radiance measurements may be estimated. The technique is a natural extension of one we previously proposed to estimate the effective data density in a profile. We test the method in a comparison of simulated satellite instruments to show that the method does indeed provide an objective summary of the spatial distribution of each instrument's information content. We propose that further extensions of the method be developed to include other more traditional data sources in a fully three-dimensional scheme. We also note that analogous and compatible methods may be used to diagnose the information content of meteorological analysis and forecast fields relative to the information contained in the covariance, at the appropriate season, of the corresponding climate fields.