It is known that the ionosphere behaves quite differently from storm-to-storm, certainly with respect to the details, and there are latitudinal, seasonal, diurnal and solar epochal effects to consider. Moreover, it makes a difference whether the onset of the storm is during the day and during the night. Empirical models are data-driven but most efforts rely upon physical insight to extrapolate results into those domains for which data are sparse or unavailable altogether. Empirical models of magnetic storm effects are elusive, a fact that is not surprising given the fact that not all theoretical questions have been fully answered.
When we speak about modeling of magnetic storm effects, it is important to understand what we use as an input and what we desire as an output. Detman and Vassiliadis  have reviewed techniques for magnetic storm forecasting, and Lundstedt  has described AI (and specifically neural network) approaches. These efforts use downstream data (i.e., solar and solar wind data) as an input to arrive at an output such as a magnetic activity index (i.e., Kp or Dst). From the perspective of telecommunications we are not uninterested in these results. But we are a bit more interested in how a given time history of A-index translates into ionospheric perturbations. A discussion of the magnetic storm forecasting (as opposed to ionospheric storm forecasting that is based upon magnetic storm attributes) is more properly a topic to be covered in Chapter 2 ("The Origins of Space Weather").
Needless to say, one can envision the marriage of A1 technology, as the basis of a model to yield A"-index time history, to an ionospheric response model (such as STORM, see below), yielding a prediction offoF2 departures. To go a step further, we would like to promote the notion that the resultant foF2 departure data set could be exploited in a number of propagation codes that require foF2 as one of the input parameters. Then we will have successfully linked "upstream" solar data sets to system performance. But that is a little ahead of the game. Let's take a brief look at the STORM model.
Fuller-Rowell et al.  have addressed the question "How does the thermosphere and ionosphere react to a geomagnetic storm?" Fuller-Rowell and his team have developed a model called STORM [Araujo-Pradere et al., 2002], The model has also been imbedded in the International Reference Ionosphere (IRI2000) [Araujo-Pradere et al., 2002], and the imbedded model shows a 28% improvement in performance during storm days when compared with the model IRI95, which does not contain a stormtime correction. (See Section 126.96.36.199.2 for a brief discussion of the realtime STORM output on the SEC website.)
Fuller-Rowell and his team have developed an empirical model of the ionospheric storm, but it is important to recognize that the model is consistent with theoretical understandings [Prolss, 1993; Fuller-Rowell, 1996]. Table 3.3 is a list of the major points underpinning the model, as indicated by Fuller-Rowell and his team at SEC.
There are many factors that can compromise the assertions implicit in the listing above. Since there are a number of processes at work, the net result depends on the blend of the physical processes. This situation is difficult to specify in a complex physical model, and virtually impossible to represent in an empirical model. While one should not expect perfection in the SEC STORM model predictions, it is currently the only model that appears to capture most of the features of interest, and is simple enough to be operationally useful.
The STORM model is now on-line, and operates in a nowcast mode. However, telecommunication specialists would benefit if certain improvements were to be implemented. First, it would be useful if the model could be modified to work in a forecast mode. It is understood that NOAA-SEC is contemplating a 12-hour forecast. Secondly, it would be useful if the model could be run with different versions of the ionospheric response filter function, especially for execution at different geomagnetic latitudes. Finally, it would be more convenient to the user if the output were organized in terms of geographic latitude and longitude, with more resolution provided.
Table 3.3: Assumptions forming the Basis for the STORM Model
□ Long-lived negative storm effects are associated with neutral composition changes
□ The so-called "compositional bulge" is from auroral heating as a result of magnetospheric input
□ The neutral air (bulge) is transported to midlatitudes by nocturnal winds that are equatorward
□ The neutral air (bulge) is brought to the dayside by earth rotation effects
□ Summer-to-Winter circulation is prevailing, sending (molecular rich) gas to middle and low latitudes in the summer hemisphere over a matter of days.
□ Poleward winds in the winter hemisphere restrict the equatorward movement of the bulge.
□ Result_l: The winter hemisphere has a net decrease in molecular species (resulting from downwelling) and this causes a positive storm.
□ Result_2: The summer hemispheric bulge introduces a net decrease in the electron concentration, and a negative storm.
Was this article helpful?