Evaluation and improvement of NCEP Global Forecast System (GFS) using satellite retrievals

Description

There are several studies that reveal the horizontal distribution of clouds from the surface and from space-borne sensors. However, the vertical distribution and internal structure of clouds have not been resolved on global scales. As such, the majority of models were evaluated by comparing radiation measurements at the top of atmosphere (TOA) and total column cloud amounts instead of layered clouds. The objective of this study is to diagnose the performance of the NCEP GFS (Global Forecast System) model using different types of satellite retrievals. Comparisons were made against the merged CloudSat and Cloud Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) data and Atmospheric InfraRed Sounder (AIRS) for July 2007 and 2008. Several cloud parameters and meteorological variables were evaluated, including Cloud Vertical Fraction (CVF), atmospheric temperature vertical profile, the lowest cloud top, base height and its thickness, and relative humidity (RH). In general, the GFS captures reasonably well the patterns of hydrometeors and follows the general trends of satellite measurements, but has large discrepancies in vertical temperature and RH profiles. GFS model results are most comparable with the observation at upper level in the troposphere but overestimated approaching the surface.