A coupled NCEP GCM/SSiB (Simplified Simple Biosphere Model) model has been developed to investigate the interactions between land surface processes and climate, in particular the interactions between land and monsoon system. The NCEP global model is a forecasting model that is also used for climate studies. A version with spectral triangular 62 truncation (T62) is used for this study. The corresponding gaussian grid for T62 is 192 by 94, roughly equivalent to 2 degrees in latitude and longitude. There are 28 unequally spaced sigma levels. About 8 levels are below 800 hpa. There are two soil layers in the original NCEP GCM, representing land surface processes. Using the original NCEP GCM and the NCEP GCM/SSiB, we have integrated the models for 4 months from May 1, 1987 to August 1, 1987, which is the boreal monsoon season.
The most substantial differences between the NCEP GCM and the NCEP GCM/SSiB are the simulations in the African monsoon, the East Asian monsoon, and the Mexican monsoon. In the original NCEP GCM, the African monsoon is too week, especially in July. The monsoon rainfall does not fully develop until August. With the coupled NCEP GCM/SSIB, the African monsoon rainfall in July is closer to the observations. In East Asia, the monsoon onset in May is too strong in the original NCEP GCM and the rainfall area approaches too far to the north; whereas the NCEP GCM/SSiB correctly simulates the monsoon s onset. In these experiments, the soil moisture and albedo are similar in the two models. (Only the NCEP soil moisture data is used as initial condition). The major differences are the surface cover conditions: one with vegetation and the other with only soil layer.
To further understand the land surface effect, a newly developed global vegetation map (UMGEOG1) and a new soil moisture data are introduced to the NCEP GCM to specify the land condition. This UMGEOG1 is developed at the Department of Geography, University of Maryland, and is based on the NOAA/NASA pathfinder AVHRR 1-km land data set. Theclassifications are derived using a decision tree classifier with training data derived from a global network of high resolution Landsat data. The land cover data sets are being validated with Landsat data as well as with regional data sets. The soil moisture data is developed atNASA/GSFC by Sud s group. It is a part of the Global Soil Wetness Project. The SSiB has been driven offline by observed and assimilated meteorological data to produce 1987 soil moisture climatology. The impacts of these two data sets have been tested to compare with the original model simulations. The physical mechanisms have also been explored.