This thesis provides a spectral optimal gridding (SOG) method to build dynamically consistent reconstructions of the global ocean temperature ranging from 5 to 5,500 meters over 63 years (1950-2012). This method utilizes empirical orthogonal functions (EOFs) derived from NASA Jet Propulsion Laboratory (JPL) non-Boussinesq ocean general circulation model (OGCM) at 1/4 degree spatial resolution and one month temporal resolution. These EOFs are computed using a reduced singular value decomposition (SVD) from the OGCM data during the time period 1958 through 1995. The SOG method is validated by the EOFs ability to accurately reconstruct the full OGCM output by using the OGCMs temperature data at places only where the observational data exists. Several SOG reconstruction examples are presented and validated against well known El Niño and La Niña events, as well as, compared to the Global Ocean Data Assimilation System (GODAS) that was developed at the U.S. National Centers for Environmental Prediction. Additionally, a full 3D SOG reconstruction of January 1998 shows a strong cold anomaly El Niño signal well below the sea surface. This suggests that an improved El Niño and possibly La Niña forecasting can be achieved using SOG reconstructions.
