Previous studies show that a coupled Eta/SSiB model is able to realistically simulate different climate scenarios. In these experiments, the model is integrated for 48 hours. The
1200Zs for June 1988 and July 1993 are used as initial conditions and the NCEP reanalysis data are used as lateral boundary conditions. The 24 hour forecast and the 48 hour forecast for June
1988 and July 1993 are averaged, respectively. However, there are two important issues related
to this kind of regional studies. First, these simulations depend on the lateral boundary conditions prescribed. The relative importance of the lateral boundary conditions in prediction
needs to be investigated. Secondly, the model is initialized every 24 hours. Therefore, the prediction for some hydrological variables, such as runoff, is mainly determined by the initial
conditions. It is desirable to explore the usage of regional Eta model for long term climate study.
To understand the importance of these initial and lateral boundary conditions (IBC) to the regional climate prediction, both reanalysis data and GCM output as IBC are used. In the 1988
drought case, the rainfall patterns of reanalysis data are generally in agreement with the observation. However, only the northern part of the drought area in central US is simulated, and
the rainfall in southeast US is much higher than the observation. The GCM simulation overestimates the precipitation. But the results in southeastern US do not have systematic wet
bias as in the reanalysis data. In the Eta/SSiB simulation with reanalysis data as the IBC, the simulated results are substantially improved. The location of the drought area and the monthly
mean precipitation are correct. The Eta/SSiB with GCM output as the IBC requires more time to adjust. Although the drought area is correctly simulated, it is much larger than the observation.
Particularly, the drought in southwest US is not found in observed data. In the 1993 wet case, the situations are very similar to the June 1988 case. The Eta/SSiB with reanalysis as IBC
exhibit the best results. The GCM simulations have serious dry bias and the Eta/SSiB with GCM as IBC has not shown any improvement. Although the land surface model provides similar fluxes to the atmosphere as in the case when reanalysis is used as IBC.
These results indicate that proper IBC is one of the key elements for regional climate studies. Although the reanalysis data
with a coupled Eta/SSiB is able to provide better prediction of the hydrological cycle, only GCM output can be used for future regional climate prediction studies. To obtain a more comprehensive
comparison, the results from recently developed NCEP global/SSiB are also used as IBC for regional study.
To tackle the second issue, a one month integration using a 1997 version of the NCEP Eta model is conducted to test the 1993 flood event. The model is dried out with no similar
variation in precipitation compared to observations. The long term predictability with Eta/SSiB for different climate events is also explored and the time scale of the predictability for the Eta
model is investigated