Utilizing NASA A-Train datasets to evaluate global models

Using collocated measurements from NASA A-Train satellites, we evaluated clouds and water vapor simulations from GEOS5-AGCM, NCAR-CAM3.5 and GDFL-AM2. To ensure consistent spatial and temporal sampling between model output and satellite measurements, 3-hourly (or 6-hourly) model outputs were interpolated onto satellite measurement locations in both space and time and with vertical averaging kernels applied. Two types of comparison were performed in this model evaluation study: Direct Comparison examines model consistency with satellite observations in terms of the global distribution, seasonal maps, tropical mean profiles, day-night difference, latitude-time section, height-time section, longitude-time section, and the response to ENSO; Conditional Sampling sorts modeled and observed parameters by large-scale meteorological variables to examine the relationships of cloud and water vapor with environmental conditions. Our analysis demonstrates the strength and weakness in model representation of clouds and moisture, and indicates possible areas of improvements.