Diagnosing diurnal convection signals in the NCEP GFS

Over many parts of the world, the diurnal signal of convection is strong. Studies have been done for the Maritime continent, India, the Amazon region, the Great Planes of North America, and Africa. Many weather and climate models have been examined for these signals and most have been shown to fail the tests. A recent study (Lee, et al 2008) showed that the NCEP GFS is one of three models examined that exhibited the nocturnal precipitation signal and provided some insights into the mechanism responsible for the signal. As a follow up to this study, we set out to examine the diurnal convection signals over other regions of the world where observational studies have shown pronounced diurnal signals.

In this study, we will examine the diurnal convection signals over the Amazons and the maritime continents using the short-range forecast archives from the NCEP GFS and the climate simulations from the NCEP CFS. The problem with the routine forecast archives is that the output intervals are coarser than we would like to study the diurnal variations. For the GFS, the precipitation bucket is emptied every three hours while the bucket for the CFS simulations is six hours. In order to further diagnose the diurnal signal, we are using a single column version of the GFS to perform short-range predictions and we will composite the forecast products over the warm seasons to compare with the operational archives.

For the oceanic region of the Maritime continents, studies have shown that diurnal convection signal also exists. We will compare the diurnal convective signal over this region from the GFS (which uses observed sea surface temperature) and the CFS (which is a coupled model of the atmosphere and the ocean) to examine the importance of the coupling.