This research project will address a problem in the initialization of global numerical weather prediction models known as "spin-up". Before the model has "spun up", there is an imbalance in the various model processes which adversely affects the humidity, precipitation, divergence and evaporation fields. The initial imbalances are due, in large part, to the data assimilation scheme. The goal of this research is to reduce the model spin-up time and to improve the precipitation forecast by improving the data assimilation scheme. The principal investigator will further refine an assimilation technique he has developed in which Newton relaxation, or nudging, is used to make the divergence and humidity fields consistent with the observed precipitation at the beginning of the forecast. This project is part of a joint program in numerical weather prediction sponsored by NSF and the National Meteorological Center (NMC) of NOAA. The principal investigator will use the NMC global forecast model for his research. If the new assimilation scheme is successful, it will be used operationally by NMC.