Published in Interface. Vol. 2, No. 2, May 1994

University of Alabama in Huntsville

by James Drake

Among the objects studied in mathematics are intriguing curves that wind and turn to visit every part of the space available to them. Like such a space-filling curve, computational research at the University of Alabama in Huntsville (UAH) extends through both a large physical volume and an appreciable number of disciplines. For now, space of another sort permits a look at just three prominent sections of the curve: aerospace engineering, studies of the lower atmosphere, and economic modeling.

Many UAH researchers use computational methods to develop better rocket systems. In the design of supersonic vehicles propulsion, aerodynamics, and aero-optics each pose the problem of determining a flow where the effect of chemical reactions on the flow may be significant. Several groups make use of finite-element models capable of modeling varying geometries and tens of reactions. They are used to test reaction schemes and turbulence models, to design wind tunnels, and to determine the optical properties of air near the windows of sensors. Many, if not most, of the projects cross the lines between disciplines. Thus, mechanical engineers use computational chemistry to give exact results that are in turn used to develop lumped schemes, suitable for use in CFD codes, for multi-step dissociation reactions.

The testing and programming of rocket guidance and propulsion systems also draw on computers. In "hardware-in-the-loop" testing, a hardware subsystem is driven by a coupled computer to simulate, in the laboratory, the total performance of a system with greater realism. Fuel-efficient trajectories for the guidance systems to follow into terrestrial, lunar, and Martian orbits are being designed with computer assistance. Other research at UAH concerns itself with the regions traversed by rockets, in the Earth's vicinity and beyond.

Atmospheric scientists at UAH both use the products of atmospheric remote sensing from satellites and develop new sensing techniques. Corrections for the effects of volcanoes and El Niño have been applied to temperatures deduced from satellite observations to give a more accurate global "thermometer." At the Center for Space Plasma and Aeronomic Research, researchers from several departments develop and apply techniques of parameter retrieval to measure important trace constituents of the atmosphere.

Numerical modeling of the atmosphere at UAH advances on a theoretical front while giving close and appropriate attention to pertinent observations. The satellite-derived temperatures mentioned above also bring a welcome standard of realism to the first major study to compare the results from a suite of atmospheric general circulation models. Measurements of stratospheric constituents made on flights by the space shuttle are being compared with a stratospheric model from Caltech. The ozone depletion sometimes seen in Antarctic winters appears to be correlated with the strength of a polar vortex. Regional atmospheric modeling is in progress to determine how winds driven by the strong seasonal cooling may affect the vortex.

In the Department of Finance, researchers are interested in correlating stock-price fluctuations to economically significant events such as new-product announcements, regulatory changes, natural disasters, etc. Stock-price models normally assume a random walk, with a volatility coefficient, beta, that varies from stock to stock, on top of a steady trend alpha, which is the same for all stocks. The usual model considers beta to be constant. UAH is testing models where beta can take discontinuous jumps. When such a jump matches actual stock price movements, the financial news is scanned for likely causes. Figure 5 tracks a correlation parameter over a 270-day period. The two horizontal axes each represent times when jumps can occur, with the vertical coordinate representing a statistical correlation function. Such a plot indicates the likelihood, at the given times, of beta jumps. Ultimately, a data base of beta values appropriate for more classes of events will be built up. Prospects for more reliable stock-price predictions seem good.