The medium range forecast model of the National Meteorological Center (NMC) has been integrated to produce winter and summer simulations. For the winter simulation, the model was initialized with the NMC analysis of 1200 UTC 15 December 1985 and integrated for 110 days to simulate the December–March period (referred to hereinafter as the DJFM simulation). For the summer simulation, the model was initialized with NMC analysis of 0000 UTC 1 May 1986 and was integrated for 110 days to simulate the May–August period (MJJA simulation). In each case, seasonally varying boundary conditions of sea surface temperature, soil moisture and sea ice were used. The computer code used was nearly identical to that used by NMC for the operational ten-day forecast during the period 16 April 1985 through 30 May 1986. Both simulations have been compared to the NMC analyses for the corresponding period.
It was found that the model climatology, defined as the average over the last 3 months of each run, is similar to that of the observed atmosphere as well as climatologies of other general circulation models. Notably, the model maintains a reasonable horizontal temperature gradient and circulation distribution, but the model is colder than observed in the troposphere nearly everywhere and cools in the lower stratosphere in the tropics and near the poles in both simulations. A detailed description of the stationary and transient features of the model circulation including both tropical and extratropical regions is also given. In addition, the model hydrological cycle, radiative balance and surface heat budget are discussed. The secondary circulations in the tropics including the zonally symmetric Hadley cell, as simulated by the model, are also compared with the observations. The upper branch of the Hadley cell appears to be only poorly simulated in DJFM.
Generally, the simulations demonstrated reasonable agreement with the observations in sea level pressure, the structure of the tropospheric zonal jets and the winter hemispheric stationary waves. The tropical rainfall is very different from climatology or surrogates for precipitation observations—such as outgoing long radiation—particularly in the excessive amount of rain produced by the model over subtropical deserts. The summer hemisphere in both simulations does not agree with observations as well as the winter hemisphere; this may be related to the tropical rainfall problem.