On the long-term secular increase in sunspot number

Correlated with the maximum amplitude (R _max) of the sunspot cycle are the sum (R _sum) and the mean (R _mean) of sunspot number over the duration of the cycle, having a correlation coefficient r equal to 0.925 and 0.960, respectively. Runs tests of R _max, R _sum, and R _mean for cycles 0–21 have probabilities of randomness P equal to 6.3, 1.2, and 9.2%, respectively, indicating a tendency for these solar-cycle related parameters to be nonrandomly distributed. The past record of these parameters can be described using a simple two-parameter secular fit, one parameter being an 8-cycle modulation (the so-called Gleissberg cycle or long period) and the other being a long-term general (linear) increase lasting tens of cycles. For each of the solar-cycle related parameters, the secular fit has an r equal to about 0.7–0.8, implying that about 50–60% of the variation in R _max, R _sum, and R _mean can be accounted for by the variation in the secular fit.Extrapolation of the two-parameter secular fit of R _max to cycle 22 suggests that the present cycle will have an R _max = 74.5 ± 49.0, where the error bar equals ± 2 standard errors; hence, the maximum amplitude for cycle 22 should be lower than about 125 when sunspot number is expressed as an annual average or it should be lower than about 130 when sunspot number is expressed as a smoothed (13-month running mean) average. The long-term general increase in sunspot number appears to have begun about the time of the Maunder minimum, implying that the 314-yr periodicity found in ancient varve data may not be a dominant feature of present sunspot cycles.     

An early estimate for the size of cycle 23

Two features are found in the modern era sunspot record (cycles 10–22: ca. 1850-present) that may prove useful for gauging the size of cycle 23, the next sunspot cycle, several years ahead of its actual onset. These features include an inferred long-term increase against time of maximum amplitude (RM, the maximum value of smoothed sunspot number for a cycle) and the apparently inherent differing natures of even- and odd-numbered sunspot cycles, especially when grouped consecutively as even-odd cycle pairs. Concerning the first feature, one finds that 6 out of the last 6 sunspot cycles have had RM 110.6 (the median value for the modern era record) and that 4 out of 6 have had RM > 150. Presuming this trend to continue, one anticipates that cycle 23 will likewise have RM 110.6 and, perhaps, RM > 150. Concerning the second feature, one finds that, when one groups sunspot cycles into consecutively paired even-odd cycles, the odd-following cycle has always been the larger cycle, 6 out of 6 times. Because cycle 22 had RM = 158.5, one anticipates that cycle 23 will have RM > 158.5. Additionally, because the average difference between RM(odd) and RM(even) for consecutively paired even-odd cycles is 40.3 units (sd = 14.2), one expects cycle 23 to have RM 162.3 (RM = 198.8 ± 36.5 at the 95% level of confidence). Further, because of the rather strong linear correlation (r = 0,959, se = 13.5) found between RM(odd) and RM(even) for consecutively paired even-odd cycles, one infers that cycle 23 should have RM 176.4 (RM = 213.9 ± 37.5 at the 95% level of confidence). Since large values of RM tend to be associated with fast rising cycles of short ascent duration and high levels of 10.7-cm solar radio flux, cycle 23 is envisioned to be potentially one of the greatest cycles of the modern era, if not the greatest.     

Predicting the maximum amplitude for the sunspot cycle from the rate of rise in sunspot number

Examined are associational aspects as they relate the maximum amplitude R _M for the sunspot cycle to the rate of rise R _t during the ascending phase, where R _M is the smoothed sunspot number at cycle maximum and R _t is the sum of the monthly mean sunspot numbers for selected 6-month intervals (t) measured from cycle onset. One finds that, prior to about 2 yr into the cycle, the rate of rise is not a reliable predictor for maximum amplitude. Only during the latter half of the ascent do the fits display strong linearity, having a coefficient of correlation r 0.9 and a standard error S _yx 20. During the first four intervals, the expected R _M and the observed R _M were found to differ by no more than 20 units of smoothed sunspot number only 25, 42, 50, and 58 % of the time; during the latter four intervals, they differed by no more than 20 units 67, 83, 92, and 100% of the time.     

PREDICTIONS OF THE FEATURES FOR SUNSPOT CYCLE 23

Smoothed monthly mean Ap indices are decomposed into two components (Ap) _c and (Ap) _n. The former is directly correlated with the current sunspot numbers, while the latter is shown to achieve its maximum correlation with the sunspot numbers after some time lag. This latter property is used to develop a method for predicting the sunspot maximum based on the observed value of (Ap) _n maximum which occurs during the preceding cycle. The value of R _M for cycle 23 predicted by this method is 149.3 ± 19.9. A method to estimate the rise time (from solar minimum to maximum) has been developed (based on analyses of Hathaway, Wilson, and Reichmann, 1994) and yields a value of 4.2 years. Using an estimate that the minimum between cycles 22 and 23 occurred in May 1996, it is predicted that the sunspot maximum for cycle 23 will occur in July 2000.     

Fitting the sunspot cycles 10–21 by a modified F-distribution density function

It has been found that sunspot cycles 10–21, represented by quarterly mean values of Zürich sunspot number, can be suitably described by the F-distribution density function provided it is modified by introducing five characteristic parameters, in order to achieve an optimal fitting of each cycle. The average cycle calculated from cycles 10–21 has been used as a basis to forecast time and magnitude of the maximum of each cycle, as a function of various numbers of the first quarterly mean values in the beginning N = 8 to 16 quarters. The standard deviations at a 99% significance level calculated from the observed values depend on N, and vary from 1.6 to 1.1 quarters and 65 to 16 units of sunspot number. A rather sufficient forecast is obtained from N = 12 quarters (with inaccuracy of ± 1.5 quarters and ± 24 units); the forecast for cycle 22 yielded, for N = 12, the values t _m = (15.4 ± 1.5) quarters ( 1990.I) and f(t _m ) = (175 ±24 units).     

Prediction of the height of solar cycle 23

Presuming a bimodal behaviour of even-odd solar cycle pairs (i.e., four modes designated asA, B, C, andD), we predict the amplitude of solar cycle 23. The bimodal properties include the dependence of maximum relative sunspot number (RM) on cycle rise time (TR) separately for odd-following and even cycles (both in two split modes), and the dependencies of odd-following on even cycles separately for cycle rise times and maximum relative sunspot numbers (each also split into two mode pairs). The procedure was first to identify the proper mode for cycle 22 (modeA), which then explicitly defines the mode for cycle 23 (modeC). The presumed mode-inherent relations were then used to estimate the rise time for cycle 23 (3.7 0.5 yr) and its maximum amplitude (195.1 17.1). A second estimate of maximum amplitude, based directly on a presumed mode-inherent relation between maximum amplitudes for even and odd cycle pairs, yields a somewhat lower value (181.3 44.3). Thus, the results of this analysis supports previous findings that cycle 23 may be one of the largest amplitude cycles ever observed.     

Did predictions of the maximum sunspot number for solar cycle No. 22 come true?

Solar cycle No. 22 which started in 1986 seems to have already passed through a maximum. The maximum annual mean sunspot number was 157 for 1989. The maximum twelve-month running average was 159, centered on July 1989. For cycle 21, the similar value was 165 centered at December 1979. Thus, cycle 22 is slightly weaker than cycle 21. Schatten and Sofia (1987) had predicted a stronger cycle 22 (170 ± 25) as compared to cycle 21 (140 ± 20). Predictions based on single variable analysis, viz., R _z (max) versus aa(min) were 165 and came true. Predictions based on a bivariate analysis, viz., R _z (max) versus aa(min) and R _z (min) were 130 and proved to be underestimates. Other techniques gave over- or underestimates.     

Extrema in Sunspot Cycle Linked to Sun's Motion

Partitions of 178.8-year intervals between instances of retrograde motion in the Sun's oscillation about the center of mass of the solar system seem to provide synchronization points for the timing of minima and maxima in the 11-year sunspot cycle. In the investigated period 1632–1990, the statistical significance of the relationship goes beyond the level P=0.001. The extrapolation of the observed pattern points to sunspot maxima around 2000.6 and 2011.8. If a further connection with long-range variations in sunspot intensity proves reliable, four to five weak sunspot cycles (R0) are to be expected after cycle 23 with medium strength (R100).     

On the use of ‘first spotless day’ as a predictor for sunspot minimum

Defining the first spotless day of a sunspot cycle as the first day without spots relative to sunspot maximum during the decline of the solar cycle, one finds that the timing of that occurrence can be used as a predictor for the occurrence of solar minimum of the following cycle. For cycle 22, the first spotless day occurred in April 1994, based on the International sunspot number index, although other indices (Boulder and American) indicated the first spotless day to have occurred earlier (September 1993). For cycles 9–14, sunspot minimum followed the first spotless day by about 72 months, having a range of 62–82 months; for cycles 15–21, sunspot minimum followed the first spotless day by about 35 months, having a range of 27–40 months. Similarly, the timing of first spotless day relative to sunspot minimum and maximum for the same cycle reveals that it followed minimum (maximum) by about 69 (18) months during cycles 9–14 and by about 90 (44) months during cycles 15–21. Accepting April 1994 as the month of first spotless day occurrence for cycle 22, one finds that it occurred 91 months into the cycle and 57 months following sunspot maximum. Such values indicate that its behavior more closely matches that found for cycles 15–21 rather than for cycles 9–14. Therefore, one infers that sunspot minimum for cycle 23 will occur in about 2–3 years, or about April 1996 to April 1997. Accepting the earlier date of first spotless day occurrence indicates that sunspot minimum for cycle 23 could come several months earlier, perhaps late 1995.The U.S. Government right to retain a non-exclusive, royalty free licence in and to any copyright is acknowledged.     

On the level of skill in predicting maximum sunspot number: A comparative study of single variate and bivariate precursor techniques

Precursor prediction techniques have generally performed well in predicting the maximum amplitude of sunspot cycles, based on cycles 10–21. Single variate methods based on minimum sunspot amplitude have reliably predicted the size of the sunspot cycle 9 out of 12 times, where a reliable prediction is defined as one having an observed maximum amplitude within the prediction interval (determined from the average error). On the other hand, single variate methods based on the size of the geomagnetic minimum have reliably predicted the size of the sunspot cycle 8 of 10 times (geomagnetic data are only available since about cycle 12). Bivariate prediction methods have, thus far, performed flawlessly, giving reliable predictions 10 out of 10 times (bivariate methods are based on sunspot and geomagnetic data). For cycle 22, single variate methods (based on geomagnetic data) suggest a maximum amplitude of about 170 ± 25, while bivariate methods suggest a maximum amplitude of about 140 ± 15; thus, both techniques suggest that cycle 22 will be of smaller maximum amplitude than that observed during cycle 19, and possibly even smaller than that observed for cycle 21. Compared to the mean cycle, cycle 22 is presently behaving as if it is a + 2.6 cycle (maximum amplitude about 225). It appears then that either cycle 22 will be the first cycle not to be reliably predicted by the combined precursor techniques (i.e., cycle 22 is an anomaly, a statistical outlier) or the deviation of cycle 22 relative to the mean cycle will substantially decrease over the next 18 months. Because cycle 22 is a large amplitude cycle, maximum smoothed sunspot number is expected to occur early in 1990 (between December 1989 and May 1990).     

The latitude of filament bands at the sunspot minimum and the activity level in the two following 11-year solar cycles

The zonal structure of the distribution of filaments is considered. The mean latitudes of two filament bands are calculated in each solar hemisphere at the minima of the sunspot cycle in the period 1924–1986: middle latitude _{2, m} and low latitude _{1, m} . It is shown that the mean latitude of the filament band _{2, m} at the minimum -m of the cycle correlates, with = 0.94, with the maximum - M sunspot area S(M) and maximum Wolf number W(M) in the succeeding solar cycle M. It is shown that the mean latitude of the low-latitude filament band _{1, m} is linearly dependent on the mean latitude filament band _{2, m + 1} at the succeeding minimum. We found a correlation of the latitude of the low-latitude filament band _{1, m} with the maximum sunspot area in the M + 1 cycle. This enables us to predict the power of two succeeding 11-year solar cycles on the basis of the latitude of filament bands at the minimum of activity, 1985–1986: W(22) - 205 ± 10, W(23) - 210 ± 10. The importance of the relationships found for theory and applied aspects is emphasized. An attempt is made to interpret the relationships physically.     

On ‘bimodality of the solar cycle’ and the duration of cycle 21

The period-growth dichotomy of the solar cycle predicts that cycle 21, the present solar cycle, will be of long duration (>133 mo), ending after July 1987. Bimodality of the solar cycle (i.e., cycles being distributed into two groups according to cycle length, based on a comparison to the mean cycle period) is clearly seen in a scatter diagram of descent versus ascent durations. Based on the well-observed cycles 8–20, a linear fit for long-period cycles (being a relatively strong inverse relationship that is significant at the 5% level and having a coefficient of determination r ² 0.66) suggests that cycle 21, having an ascent of 42 mo, will have a descent near 99 mo; thus, cycle duration of about 141 mo is expected. Like cycle 11, cycle 21 occurs on the downward envelope of the sunspot number curve, yet is associated with an upward first difference in amplitude. A comparison of individual cycle, smoothed sunspot number curves for cycles 21 and 11 reveals striking similarity, which suggests that if, indeed, cycle 21 is a long-period cycle, then it too may have an extended tail of sustained, low, smoothed sunspot number, with cycle 22 minimum occurring either in late 1987 or early 1988.     

Polar faculae and sunspot cycles

The monthly number of polar faculae of the Sun were determined from white-light images at spectral band (eff) = (4100 ± 200) Å obtained at the Kislovodsk Solar Station during 1960–1994. Corrected monthly numbers were obtained with the help of the visibility function. The level of polar activity larger than 1 above the monthly running mean was calculated, and the relation between the polar faculae and sunspot cycle was studied. We confirmed earlier results (Makarov and Makarova, 1987) that the monthly number of polar faculae, NPF _m (t) correlates with the monthly sunspot area A _m (Sp)(t + T) with a time shift T 6 yr. The new polar faculae cycle began in the middle of 1991. Peculiarities of the first part of sunspot cycle 23 are discussed.Guest scientist with the University of Arizona and Zetetic Institute. Tucson, Arizona 85719, U.S.A.     

The G–O Rule and Waldmeier Effect in the Variations of the Numbers of Large and Small Sunspot Groups

We have analyzed the combined Greenwich and Solar Optical Observing Network (SOON) sunspot group data during the period of 1874??C?2011 and determined variations in the annual numbers (counts) of the small (maximum area A _M<100 millionth of solar hemisphere, msh), large (100??A _M<300?msh), and big (A _M??300?msh) spot groups. We found that the amplitude of an even-numbered cycle of the number of large groups is smaller than that of its immediately following odd-numbered cycle. This is consistent with the well known Gnevyshev and Ohl rule (G?CO rule) of solar cycles, generally described by using the Zurich sunspot number (R _Z). During cycles 12??C?21 the G?CO rule holds good for the variation in the number of small groups also, but it is violated by cycle pair (22, 23) as in the case of R _Z. This behavior of the variations in the small groups is largely responsible for the anomalous behavior of R _Z in cycle pair (22, 23). It is also found that the amplitude of an odd-numbered cycle of the number of small groups is larger than that of its immediately following even-numbered cycle. This might be called the ??reverse G?CO rule??. In the case of the number of the big groups, both cycle pairs (12, 13) and (22, 23) violated the G?CO rule. In many cycles the positions of the peaks of the small, large, and big groups are different, and considerably differ with respect to the corresponding positions of the R _Z peaks. In the case of cycle?23, the corresponding cycles of the small and large groups are largely symmetric/less asymmetric (the Waldmeier effect is weak/absent) with their maxima taking place two years later than that of R _Z. The corresponding cycle of the big groups is more asymmetric (strong Waldmeier effect) with its maximum epoch taking place at the same time as that of R _Z.     

On the average rate of growth in sunspot number and the size of the sunspot cycle

The average rate of growth during the ascending portion of the sunspot cycle, defined here as the difference in smoothed sunspot number values between elapsed time (in months) t and sunspot minimum divided by t, is shown to correlate (r 0.78) with the size of the sunspot cycle, especially for t 18 months. Also, the maximum value of the average rate of growth is shown to highly correlate (r = 0.98) with the size of the cycle. Based on the first 18 months of the cycle, cycle 22 is projected to have an R(M) = 186.0 ± 27.2 (at the ± 1 level), and based on the first 24 months of the cycle, it is projected to have an R(M) = 201.0 ± 20.1 (at the ± 1 level). Presently, the average rate of growth is continuing to rise, having a value of about 4.5 at 24 months into the cycle, a value second only to that of cycle 19 (4.8 at t = 24 and a maximum value of 5.26 at t = 27). Using 4.5 as the maximum value of the average rate of growth for cycle 22, a lower limit can be estimated for R(M); namely R(M) for cycle 22 is estimated to be 164.0 (at the 97.5% level of confidence). Thus, these findings are consistent with the previous single variate predictions that project R(M) for cycle 22 to be one of the greatest on record, probably larger than cycle 21 (164.5) and near that of cycle 19 (201.3).     

Photoelectric polarimetry of a dark unipolar sunspot

Photoelectric polarisation measurements in a stable sunspot (type H) with a particularly dark umbra carried out with the Capri magnetograph have been evaluated in terms of Unno's (1956) theory to give the value and direction of the magnetic field vector. A linear increase of the inclination angle with distance from the spot centre up tor 1.2R _s results, as originally found by Hale and Nicholson (1938) with a different procedure. The field strength decreases from the maximum value (about 3300) to about 15% at the penumbral border, still continuously decreasing outside the spot. The projected field direction deviates considerably from the radial symmetry in several parts of the spot region, but it is in good agreement with that of the overlaying chromospheric structures.Mitteilungen aus dem Fraunhofer Institut Nr. 97.     

Long-Term Variations in Solar Differential Rotation and Sunspot Activity

The solar equatorial rotation rate, determined from sunspot group data during the period 1879–2004, decreased over the last century, whereas the level of activity has increased considerably. The latitude gradient term of the solar rotation shows a significant modulation of about 79 year, which is consistent with what is expected for the existence of the Gleissberg cycle. Our analysis indicates that the level of activity will remain almost the same as the present cycle during the next few solar cycles (i.e., during the current double Hale cycle), while the length of the next double Hale cycle in sunspot activity is predicted to be longer than the current one. We find evidence for the existence of a weak linear relationship between the equatorial rotation rate and the length of sunspot cycle. Finally, we find that the length of the current cycle will be as short as that of cycle 22, indicating that the present Hale cycle may be a combination of two shorter cycles. Presently working for the Mt. Wilson Solar Archive Digitization Project at UCLA.     

Preferred longitudes of sunspot groups and high-speed solar wind streams: Evidence for a ‘solar memory’

Correlation analysis of the mean longitude distribution of sunspot groups (taken from the Greenwich Photoheliographic Results) and high-speed solar wind streams (inferred from the C9 index for geomagnetic disturbances) with the Bartels rotation period P = 27.0 days shows anti-correlation for individual cycles.In particular, the longitudes of post-maximum stable streams of cycle 18 and 19 are well anticorrelated with the preferred longitudes of sunspot groups during the maximum activity periods of these cycles. This is further analyzed using the daily Zürich sunspot number, R, between 1932 and 1980, which reveals a conspicuous similarity of cycle 18 and 19 as well as cycle 20 and 21.We conclude that there is a solar memory for preferred longitudes of activity extending at least over one, probably two cycles (i.e. one magnetic cycle of 22 years). We conjecture that this memory extends over longer intervals of time as a long-term feature of solar activity.     

Additional measurements of the high-latitude sunspot rotation rate

We report measurements of the sunspot rotation rate at high sunspot latitutdes for the years 1966–1968. Ten spots at ¦latitude¦ 28 deg were found in our Mees Solar Observatory H patrol records for this period that are suitable for such a study. On the average we find a sidereal rotation rate of 13.70 ± 0.07 deg day^-1 at 31.05 ± 0.01 deg. This result is essentially the same as that obtained by Tang (1980) for the succeeding solar cycle, and significantly larger than Newton and Nunn's (1951) results for the 1934–1944 cycle. Taken together, the full set of measurements in this latitude regime yield a rotation rate in excellent agreement with the result =14^°.37₇–2^°.77 sin², derived by Newton and Nunn from recurrent spots predominatly at lower latitudes throughout the six cycles from 1878–1944.Summer Research Assistant.     

Lags and Hysteresis Loops of Cosmic Ray Intensity Versus Sunspot Numbers: Quantitative Estimates for Cycles 19 – 23 and a Preliminary Indication for Cycle 24

Hysteresis plots between cosmic-ray (CR) intensity (recorded at the Climax station) and sunspot relative number R _Z show broad loops in odd cycles (19, 21, and 23) and narrow loops in even cycles (20 and 22). However, in the even cycles, the loops are not narrow throughout the whole cycle; around the sunspot-maximum period, a broad loop is seen. Only in the rising and declining phases, the loops are narrow in even cycles. The CR modulation is known to have a delay with respect to R _Z, and the delay was believed to be longer in odd cycles (19, 21, and 23; about 10 months) than the delay in even cycles (20 and 22; about 3?–?5 months). When this was reexamined, it was found that the delays are different during the sunspot-minimum periods (2, 6, and 14 months for odd cycles and 7 and 9 months for even cycles) and sunspot-maximum periods (0, 4, and 7 months for odd cycles and 5 and 8 months for even cycles). Thus, the differences between odd and even cycles are not significant throughout the whole cycle. In the recent even cycle 24, hysteresis plots show a preliminary broadening near the sunspot maximum, which occurred recently (February 2012). The CR level (recorded at Newark station) is still high in 2013, indicating a long lag (exceeding 10 months) with respect to the sunspot maximum.