A Retrospective Analysis
of the Pioneering Data Assimilation Experiments
With the Mintz-Arakawa General Circulation Model


Milton Halem, Jules Kouatchou and Andrea Hudson


The Nimbus 3 satellite, launched in 1969, carried a scanning infra-red sounder that provided the first global atmospheric temperature profiles claiming accuracies comparable to that obtained from radiosondes in clear regions. Shortly thereafter, Charney, Halem, and Jastrow published in J.A.S in 1969 [1] a series of numerical experiments based on a two-level General Circulation Model (GCM) generously provided by Professors Y. Mintz and A. Arakawa. The purpose of the experiment was to test the Charney conjecture that one could infer the large scale wind fields at all latitudes from a continuous history of the complete temperature fields.

The results of those experiments supported the conjecture by showing that, not only the winds, but the complete state of the atmosphere, including sea level pressure, could be determined at all latitudes. The accuracy of the inferred atmospheric states were shown to depend proportionally on the accuracy of the temperature fields and the frequencies of global coverage. These simulation studies, and those following later by these and other authors helped to usher in a new era of remote sensing observing systems for weather forecasting that was to become operational for the next three decades. Unfortunately, operational sounders never realized the potential forecast impacts of those simulation studies for several reasons, not the least being that sounding accuracies were more than twice the assumed errors in the simulation studies.

In less than two years from this symposium, if the NASA Earth Observing System (EOS) mission holds to schedule, another promising era of remote sensing sounder capabilities will unfold. The EOS PM satellite will carry advanced infra-red and microwave sounders expected to produce significantly more accurate temperature profiles that should meet the desired range of 1^o C temperature requirements described in reference [1].

Since 1969, GCMs have evolved significantly as computers have greatly increased in speed and memory over the past three decades. This has allowed major increases in horizontal and vertical resolution. In addition, scientists like Professor Arakawa and many of his students and colleagues have made substantial improvements in the fluid dynamic formulations and cloud physics parameterizations. Other scientists have also contributed to improvements in the GCM by incorporating many other physical processes such as shortwave and longwave radiations, boundary layer theory, turbulent approximations, surface processes and other parametrizations. Thus, in preparation for the new EOS era and for the purpose of this symposium, a retrospective series of numerical experiments were conducted to validate again the Charney conjecture within the context of contemporary state-of-the-art GCMs.

In the current experiments, we employ a recent version of the Goddard Earth Observing System (GEOS) GCM developed by the Data Assimilation Office at NASA Goddard Space Flight Center [2]. The resolution version used in this study is the 4^o latitude by 5^o longitude by 20 vertical levels (4^o x 5^o x 20) extended to 20mb. The experiments were designed to closely duplicate those of [1]. Results of these experiments for temperature states with 0^o, 1^o, and 2.5^o root-mean-square errors are presented. The results with a contemporary model largely confirm that a history of accurate temperature profiles can be utilized to infer the complete extra tropical state of the atmosphere but only slight improvement in the tropics, contrary to earlier findings.


[1] J. Charney, M. Halem and R. Jastrow, Use of Incomplete Historical Data to Infer the Present State of the Atmospher, Part 2, Journ. Atmos. Sci., Vol. 26, No. 5, pp. 1160-1163, September 1969.

[2] L. Takacs, A. Molod and T. Wang, Documentation of the Goddard Earth Observing System GEOS General Circulation Model - Version 1, NASA Technical Memorandum, No. 104606, September 1994.