## Prediction of the Sunspot Cycle

All techniques used for long-term prediction of sunspots have significant error bars, and, because of this, the value of long-term predictions is questionable for any detailed propagation analysis or a meaningful evaluation of telecommunication impairments. While system performance assessment and system design factors both depend upon solar activity, long-term predictions are of more value in the latter instance. Other matters such as station-keeping, orbital decay probability, and spacecraft charging, which depend upon sunspot number at some level, can influence design parameters.

360 j

360 j

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Figure 2-7: The variation of the daily sunspot number from January 1881 to January 1989. Curve is adapted from EOS Trans, AGU, vol.70, No. 32, 1989.

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Figure 2-7: The variation of the daily sunspot number from January 1881 to January 1989. Curve is adapted from EOS Trans, AGU, vol.70, No. 32, 1989.

These parameters might include the fade margins and link parameters that must be imbedded in the engineering of specified telecommunication systems. Predictions might also be useful for estimating the anticipated level of impairments years in advance by associating outages incurred in the past under the same conditions as predicted. Long-term planning for military exercises could exploit good information regarding the 11-year cycle. But how good are we at doing this?

A number of methods have been used to evaluate future sunspot cycles and no method is very precise. There are several metrics to consider, including the accuracy in prediction of the maximum amplitude, the time of the maximum, and the general shape of the cycle. For the solar cycle 23, workers at NASA have examined this problem and have predicted a maximum value of 154 ± 21, whereas the maximum of the running average was observed to be ~ 125, slightly below the margin of error. A panel of experts organized by NOAA-SEC, and sponsored by NASA, produced a report entitled Solar Cycle 23 Project: Summary of Panel Findings [Joselyn et al., 1996]. Without going into a description of the methods, Table 2-3 gives the range of values for the various techniques, and the consensus prediction, according to the panel.

I 1947 1953 1959 1965 1972 1978 1984 1990 1997 2003

Figure 2-8: Five solar cycle pattern of the 10.7 cm solar flux from 1947-2003. This data set is derived from the SOLAR2000 model, which was developed by Space Environment Technologies (SET). Figure is provided courtesy of Kent Tobiska of SET.

I 1947 1953 1959 1965 1972 1978 1984 1990 1997 2003

Figure 2-8: Five solar cycle pattern of the 10.7 cm solar flux from 1947-2003. This data set is derived from the SOLAR2000 model, which was developed by Space Environment Technologies (SET). Figure is provided courtesy of Kent Tobiska of SET.

It is evident from Table 2-3 that the timing of the sunspot cycle maximum is predicted with acceptable accuracy, whereas the observed value of the amplitude (i.e., 12-month running average of spots) is outside (i.e., below) the predicted range. Of all the methods cited, the "full" climatology technique appears to be best, followed by the neural network approach. This brief discussion serves to illustrate the point that even a panel of experts can have difficulty in assembling a prediction for a future sunspot cycle before its onset.

0 0