In a Four Dimensional Data Assimilation (FDDA) scheme, Newtonian relaxation or "nudging" of a models fields to the observations is added into the model equations to reduce error growth in amplitude and phase. As the data is ingested into the model over a period of time, each observation is weighted according to its position to a relative grid box. This study is performed with the Penn State University/National Center for Atmospheric Research Limited-Area Mesoscale Model (PSU/NCAR MM5). Currently, the weighting spatial variability of the inserted model observations can be further examined. In addition, the temporal weighting function currently assumes a time period window in which the observation is accepted as perfect with a linearly decreasing downward weighting. The design of this weighting function comes only from past model observation.

This project seeks to discover the sensitivity of the temporal and spatial weighting functions. In particular, the effectiveness of the data assimilation time and spatial periods will be varied to narrow the optimal weighting time and spatial period for the data that is used for FDDA. The shape of the weighting functions is examined in an attempt to develop and justify a more ideal shape for the weighting functions for data assimilation spatially and temporally. Finally, experiments proceed in research mode in which the ingestion of new data causes the MM5 to rerun a time period of data assimilation and approach the known existing data that was already incorporated into the model. Improvements in error are sought after by adjusting the weighting function to include data assimilation before the time the observation is received in a realtime situation.