Search

Short-term climate prediction based on a regional climate dynamical model heavily depends on atmospheric forcing and initial soil moisture state. In this study, the Weather Research and Forecasting (WRF) Model with different nudging schemes is used for approximate 2-yr simulations for investigating the importance of soil variables in seasonal temperature and precipitation simulations. The results show that the improvement of seasonal climate simulation (precipitation and air temperature) is more evident in the experiment of assimilating both soil and atmospheric variables than that in the experiments of assimilating atmospheric variables only. Further investigation of the impact of indirectly assimilating soil moisture on precipitation prediction with an indirect soil nudging (ISN) scheme shows that the precipitation reproducibility in summer is better than that in winter, and the effect of ISN is particularly prominent in the region where seasonal precipitation exceeds 200 mm. Moreover, statistical results also illustrate that initial soil moisture plays a crucial role in seasonal precipitation forecasts because of its slowly evolving nature, and its effect is more distinct in semiarid and semihumid regions than in arid and humid regions. The effects of indirectly assimilating soil moisture on precipitation can last two and three months in semiarid and semihumid areas, respectively. [ABSTRACT FROM AUTHOR]

Outputs from two ensembles of atmospheric model simulations for 1951–98 define the influence of “realistic” land surface wetness on seasonal precipitation predictability in boreal summer. The ensembles consist of one forced with observed sea surface temperatures (SSTs) and the other forced with realistic land surface wetness as well as SSTs. Predictability was determined from correlations between the time series of simulated and observed precipitation. The ratio of forced variance to total variance determined potential predictability. Predictability occurred over some land areas adjacent to tropical oceans without land wetness forcing. On the other hand, because of the chaotic nature of the atmosphere, considerable parts of the land areas of the globe did not even show potential predictability with both land wetness and SST forcings. The use of land wetness forcing enhanced predictability over semiarid regions. Such semiarid regions are generally characterized by a negative correlation between fluxes of latent heat and sensible heat from the land surface, and are “water-regulating” areas where soil moisture plays a governing role in land–atmosphere interactions. Actual seasonal prediction may be possible in these regions if slowly varying surface conditions can be estimated in advance. In contrast, some land regions (e.g., south of the Sahel, the Amazon, and Indochina) showed little predictability despite high potential predictability. These regions are mostly characterized by a positive correlation between the surface fluxes, and are “radiation-regulating” areas where the atmosphere plays a leading role. Improvements in predictability for these regions may require further improvements in model physics. [ABSTRACT FROM AUTHOR]