Studies of global climate change, interannual and decadal variability, and their effects on extreme events such as floods, droughts, and heat waves require a detailed, daily circulation dataset for the past 100+ years. While monthly global analyses of sea level pressure (SLP), temperature, and precipitation are commonly available back to 1875, and newly-available NCEP/NCAR four-times daily global reanalyses are available back to 1948, the only daily analyses available before 1948 are hand-drawn northern hemisphere synoptic SLP maps. These maps contain inhomogeneities, discontinuities, and errors, especially during World War I and the 1930's US drought period. While prior to 1948, surface station and marine pressure data are available, there are no radiosondes to provide vertical atmospheric structure information to a modern data assimilation system (DAS). It is widely believed that without upper-air observations, DAS will fail to provide a useful estimate of the synoptic state of the atmosphere. Therefore, 3-dimensional reanalyses of this period have not been attempted. However, surface pressure provides an integrated measure of the atmospheric vertical column and may provide sufficient information for DAS to reanalyze the surface, as well as the lower tropospheric circulation.

We have investigated the feasibility of using surface pressure, together with other available surface variables, to create a homogeneous, high-quality daily analysis of the surface and lower tropospheric circulation back to the late 19th century. We have performed parallel assimilation experiments for November 1999-February 2000 using surface data at observational densities expected for 1900. Two systems were compared, statistical optimal interpolation (OI) and the NCEP operational DAS. We have found that OI, using climatology as first-guess, is extremely competitive with the NCEP operational DAS in the northern hemisphere, with both producing high quality 4-times daily surface pressure fields in our test period. In the Tropics, however, the OI is superior. In addition to surface pressure, the DAS is able to produce realistic lower-tropospheric analyses through the model's dynamics. The results suggest that historical surface and lower tropospheric reanalysis is extremely feasible and may benefit from more advanced data assimilation techniques such as the ensemble Kalman filter and smoother.