Abstract
Radiative effects of clouds on 8-day northern hemispheric prediction are investigated in the real data prediction experiment using several versions of a 4-level northern hemispheric primitive equation model. They are C0-model (0% cloud cover at all levels), C1-model (100% cloud cover at all levels) and CP-model (cloud cover at each level is determined from the predicted relative humidity).
Main differences of 8-day northern hemispheric predictions between C0 and C1 are found in the vertical profile of average temperature over the whole horizontal area both in summer and in winter, and in the latitudinal gradient of both temperature at lower troposphere and 500mb height in winter.
The differences can be interpreted to stem from the following three effects of clouds through radiation; greenhouse effect through infrared radiation, albedo effect through solar radiation and ground effect through sensible and latent heat flux at the ground surface. The greenhouse effect works to warm the air under the cloud and to cool the air above the cloud, compared with C0-model. The albedo effect works to cool the air under the cloud and to warm the air above the cloud, compared with C0-model. Since the upward sensible and latent heat flux at the ground surface are produced by warming of the ground surface due to absorption of solar radiation, the ground effect works to cool the air under the cloud compared with C0-model.
For cloudy cases the greenhouse effect tends to make the troposphere unstable, while the albedo and the ground effect tend to make the troposphere stable, compared with clear cases. The albedo effect is limited in the daytime, and the ground effect is limited in the daytime and over the land area. Therefore, the greenhouse effect of clouds is the most dominant effect in interpreting the differences of the 8-day northern hemispheric predictions by C0-and C1-model.
